twitter-sentiment-analysis-with-bert-and-roberta-t.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":1472453,"sourceType":"datasetVersion","datasetId":863934}],"dockerImageVersionId":30153,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Twitter Sentiment Analysis with BERT and roBERTa Transformers","metadata":{}},{"cell_type":"code","source":"# traverse files in /kaggle/input\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:41.049301Z","iopub.execute_input":"2024-09-23T19:01:41.049879Z","iopub.status.idle":"2024-09-23T19:01:41.056649Z","shell.execute_reply.started":"2024-09-23T19:01:41.049841Z","shell.execute_reply":"2024-09-23T19:01:41.055874Z"},"trusted":true},"execution_count":5,"outputs":[{"name":"stdout","text":"/kaggle/input/covid-19-nlp-text-classification/Corona_NLP_test.csv\n/kaggle/input/covid-19-nlp-text-classification/Corona_NLP_train.csv\n","output_type":"stream"}]},{"cell_type":"code","source":"# custom confusion matrix\ndef conf_matrix(y, y_pred, title):\n    fig, ax =plt.subplots(figsize=(5,5))\n    labels=['Negative', 'Neutral', 'Positive']\n    ax=sns.heatmap(confusion_matrix(y, y_pred), annot=True, cmap=\"Blues\", fmt='g', cbar=False, annot_kws={\"size\":25})\n    plt.title(title, fontsize=20)\n    ax.xaxis.set_ticklabels(labels, fontsize=17) \n    ax.yaxis.set_ticklabels(labels, fontsize=17)\n    ax.set_ylabel('Test', fontsize=20)\n    ax.set_xlabel('Predicted', fontsize=20)\n    plt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:41.064702Z","iopub.execute_input":"2024-09-23T19:01:41.064891Z","iopub.status.idle":"2024-09-23T19:01:41.073696Z","shell.execute_reply.started":"2024-09-23T19:01:41.064857Z","shell.execute_reply":"2024-09-23T19:01:41.072945Z"},"trusted":true},"execution_count":7,"outputs":[]},{"cell_type":"code","source":"# load train and test data into dataframes\nimport pandas as pd\ndf = pd.read_csv('../input/covid-19-nlp-text-classification/Corona_NLP_train.csv',encoding='ISO-8859-1')\ndf_test = pd.read_csv('../input/covid-19-nlp-text-classification/Corona_NLP_test.csv')","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.074592Z","iopub.execute_input":"2024-09-23T19:01:41.075653Z","iopub.status.idle":"2024-09-23T19:01:41.403146Z","shell.execute_reply.started":"2024-09-23T19:01:41.075613Z","shell.execute_reply":"2024-09-23T19:01:41.402516Z"},"trusted":true},"execution_count":8,"outputs":[]},{"cell_type":"code","source":"df.head()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.405428Z","iopub.execute_input":"2024-09-23T19:01:41.406126Z","iopub.status.idle":"2024-09-23T19:01:41.424639Z","shell.execute_reply.started":"2024-09-23T19:01:41.406081Z","shell.execute_reply":"2024-09-23T19:01:41.423901Z"},"trusted":true},"execution_count":9,"outputs":[{"execution_count":9,"output_type":"execute_result","data":{"text/plain":"   UserName  ScreenName   Location     TweetAt  \\\n0      3799       48751     London  16-03-2020   \n1      3800       48752         UK  16-03-2020   \n2      3801       48753  Vagabonds  16-03-2020   \n3      3802       48754        NaN  16-03-2020   \n4      3803       48755        NaN  16-03-2020   \n\n                                       OriginalTweet           Sentiment  \n0  @MeNyrbie @Phil_Gahan @Chrisitv https://t.co/i...             Neutral  \n1  advice Talk to your neighbours family to excha...            Positive  \n2  Coronavirus Australia: Woolworths to give elde...            Positive  \n3  My food stock is not the only one which is emp...            Positive  \n4  Me, ready to go at supermarket during the #COV...  Extremely Negative  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>UserName</th>\n      <th>ScreenName</th>\n      <th>Location</th>\n      <th>TweetAt</th>\n      <th>OriginalTweet</th>\n      <th>Sentiment</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>3799</td>\n      <td>48751</td>\n      <td>London</td>\n      <td>16-03-2020</td>\n      <td>@MeNyrbie @Phil_Gahan @Chrisitv https://t.co/i...</td>\n      <td>Neutral</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>3800</td>\n      <td>48752</td>\n      <td>UK</td>\n      <td>16-03-2020</td>\n      <td>advice Talk to your neighbours family to excha...</td>\n      <td>Positive</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3801</td>\n      <td>48753</td>\n      <td>Vagabonds</td>\n      <td>16-03-2020</td>\n      <td>Coronavirus Australia: Woolworths to give elde...</td>\n      <td>Positive</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>3802</td>\n      <td>48754</td>\n      <td>NaN</td>\n      <td>16-03-2020</td>\n      <td>My food stock is not the only one which is emp...</td>\n      <td>Positive</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>3803</td>\n      <td>48755</td>\n      <td>NaN</td>\n      <td>16-03-2020</td>\n      <td>Me, ready to go at supermarket during the #COV...</td>\n      <td>Extremely Negative</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"df.info()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.425659Z","iopub.execute_input":"2024-09-23T19:01:41.425842Z","iopub.status.idle":"2024-09-23T19:01:41.463017Z","shell.execute_reply.started":"2024-09-23T19:01:41.425818Z","shell.execute_reply":"2024-09-23T19:01:41.462213Z"},"trusted":true},"execution_count":10,"outputs":[{"name":"stdout","text":"<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 41157 entries, 0 to 41156\nData columns (total 6 columns):\n #   Column         Non-Null Count  Dtype \n---  ------         --------------  ----- \n 0   UserName       41157 non-null  int64 \n 1   ScreenName     41157 non-null  int64 \n 2   Location       32567 non-null  object\n 3   TweetAt        41157 non-null  object\n 4   OriginalTweet  41157 non-null  object\n 5   Sentiment      41157 non-null  object\ndtypes: int64(2), object(4)\nmemory usage: 1.9+ MB\n","output_type":"stream"}]},{"cell_type":"code","source":"# convert TweetAt column to pandas datetime format\ndf['TweetAt'] = pd.to_datetime(df['TweetAt'])","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.464057Z","iopub.execute_input":"2024-09-23T19:01:41.464244Z","iopub.status.idle":"2024-09-23T19:01:41.478152Z","shell.execute_reply.started":"2024-09-23T19:01:41.464221Z","shell.execute_reply":"2024-09-23T19:01:41.477626Z"},"trusted":true},"execution_count":11,"outputs":[]},{"cell_type":"code","source":"# remove duplicate tweets\ndf.drop_duplicates(subset='OriginalTweet',inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.478973Z","iopub.execute_input":"2024-09-23T19:01:41.479149Z","iopub.status.idle":"2024-09-23T19:01:41.516454Z","shell.execute_reply.started":"2024-09-23T19:01:41.479127Z","shell.execute_reply":"2024-09-23T19:01:41.515892Z"},"trusted":true},"execution_count":12,"outputs":[]},{"cell_type":"code","source":"df.info()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:41.517572Z","iopub.execute_input":"2024-09-23T19:01:41.517865Z","iopub.status.idle":"2024-09-23T19:01:41.544565Z","shell.execute_reply.started":"2024-09-23T19:01:41.517824Z","shell.execute_reply":"2024-09-23T19:01:41.543556Z"},"trusted":true},"execution_count":13,"outputs":[{"name":"stdout","text":"<class 'pandas.core.frame.DataFrame'>\nInt64Index: 41157 entries, 0 to 41156\nData columns (total 6 columns):\n #   Column         Non-Null Count  Dtype         \n---  ------         --------------  -----         \n 0   UserName       41157 non-null  int64         \n 1   ScreenName     41157 non-null  int64         \n 2   Location       32567 non-null  object        \n 3   TweetAt        41157 non-null  datetime64[ns]\n 4   OriginalTweet  41157 non-null  object        \n 5   Sentiment      41157 non-null  object        \ndtypes: datetime64[ns](1), int64(2), object(3)\nmemory usage: 2.2+ MB\n","output_type":"stream"}]},{"cell_type":"code","source":"# count tweets per day\ntweets_per_day = df['TweetAt'].dt.strftime('%m-%d').value_counts().sort_index().reset_index(name='counts')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:41.545961Z","iopub.execute_input":"2024-09-23T19:01:41.546207Z","iopub.status.idle":"2024-09-23T19:01:41.843727Z","shell.execute_reply.started":"2024-09-23T19:01:41.546177Z","shell.execute_reply":"2024-09-23T19:01:41.842955Z"},"trusted":true},"execution_count":14,"outputs":[]},{"cell_type":"code","source":"# plot tweets per day count\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nplt.figure(figsize=(20,5))\nax = sns.barplot(x='index', y='counts', data=tweets_per_day,edgecolor = 'black',ci=False, palette='Blues_r')\nplt.title('Tweets count by date')\nplt.yticks([])\nax.bar_label(ax.containers[0])\nplt.ylabel('count')\nplt.xlabel('')\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:41.844827Z","iopub.execute_input":"2024-09-23T19:01:41.845053Z","iopub.status.idle":"2024-09-23T19:01:42.576165Z","shell.execute_reply.started":"2024-09-23T19:01:41.845025Z","shell.execute_reply":"2024-09-23T19:01:42.575422Z"},"trusted":true},"execution_count":15,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x360 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# count tweets per country\ntweets_per_country = df['Location'].value_counts().loc[lambda x : x > 100].reset_index(name='counts')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:42.577213Z","iopub.execute_input":"2024-09-23T19:01:42.577416Z","iopub.status.idle":"2024-09-23T19:01:42.591641Z","shell.execute_reply.started":"2024-09-23T19:01:42.577391Z","shell.execute_reply":"2024-09-23T19:01:42.590947Z"},"trusted":true},"execution_count":16,"outputs":[]},{"cell_type":"code","source":"# plot tweets per country count\nplt.figure(figsize=(15,6))\nax = sns.barplot(x='index', y='counts', data=tweets_per_country,edgecolor = 'black',ci=False, palette='Spectral')\nplt.title('Tweets count by country')\nplt.xticks(rotation=70)\nplt.yticks([])\nax.bar_label(ax.containers[0])\nplt.ylabel('count')\nplt.xlabel('')\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:01:42.592744Z","iopub.execute_input":"2024-09-23T19:01:42.592974Z","iopub.status.idle":"2024-09-23T19:01:43.019402Z","shell.execute_reply.started":"2024-09-23T19:01:42.592934Z","shell.execute_reply":"2024-09-23T19:01:43.018684Z"},"trusted":true},"execution_count":17,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1080x432 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# data cleaning -> just keep columns OriginalTweet (raw tweet) and target Sentiment\ndf = df[['OriginalTweet','Sentiment']]","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:43.022420Z","iopub.execute_input":"2024-09-23T19:01:43.022670Z","iopub.status.idle":"2024-09-23T19:01:43.028844Z","shell.execute_reply.started":"2024-09-23T19:01:43.022641Z","shell.execute_reply":"2024-09-23T19:01:43.028055Z"},"trusted":true},"execution_count":18,"outputs":[]},{"cell_type":"code","source":"df_test = df_test[['OriginalTweet','Sentiment']]","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:43.029980Z","iopub.execute_input":"2024-09-23T19:01:43.030244Z","iopub.status.idle":"2024-09-23T19:01:43.039609Z","shell.execute_reply.started":"2024-09-23T19:01:43.030210Z","shell.execute_reply":"2024-09-23T19:01:43.038921Z"},"trusted":true},"execution_count":19,"outputs":[]},{"cell_type":"code","source":"# data cleaning -> custom functions to clean tweets content\n\nimport re, string, emoji, nltk\n\n# remove emojis\ndef strip_emoji(text):\n    return re.sub(emoji.get_emoji_regexp(), r\"\", text)\n\n# remove punctuations, links, mentions, new line characters, non utf8/ascii\ndef strip_all_entities(text): \n    text = text.replace('\\r', '').replace('\\n', ' ').lower()\n    text = re.sub(r\"(?:\\@|https?\\://)\\S+\", \"\", text)\n    text = re.sub(r'[^\\x00-\\x7f]',r'', text)\n    banned_list= string.punctuation + 'Ã'+'±'+'ã'+'¼'+'â'+'»'+'§'\n    table = str.maketrans('', '', banned_list)\n    text = text.translate(table)\n    return text\n\n# remove hashtags at the end of the sentence\n# keep hashtags in the middle of the sentence by removing just the # symbol\ndef clean_hashtags(tweet):\n    new_tweet = \" \".join(word.strip() for word in re.split('#(?!(?:hashtag)\\b)[\\w-]+(?=(?:\\s+#[\\w-]+)*\\s*$)', tweet))\n    new_tweet2 = \" \".join(word.strip() for word in re.split('#|_', new_tweet))\n    return new_tweet2\n\n# remove special characters such as &, $\ndef filter_chars(a):\n    sent = []\n    for word in a.split(' '):\n        if ('$' in word) | ('&' in word):\n            sent.append('')\n        else:\n            sent.append(word)\n    return ' '.join(sent)\n\n# remove multiple spaces\ndef remove_mult_spaces(text):\n    return re.sub(\"\\s\\s+\" , \" \", text)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:43.040790Z","iopub.execute_input":"2024-09-23T19:01:43.041258Z","iopub.status.idle":"2024-09-23T19:01:43.531695Z","shell.execute_reply.started":"2024-09-23T19:01:43.041221Z","shell.execute_reply":"2024-09-23T19:01:43.530969Z"},"trusted":true},"execution_count":20,"outputs":[]},{"cell_type":"code","source":"# apply tweet content cleaning functions to training data\ntexts_new = []\nfor t in df.OriginalTweet:\n    texts_new.append(remove_mult_spaces(filter_chars(clean_hashtags(strip_all_entities(strip_emoji(t))))))","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:01:43.532928Z","iopub.execute_input":"2024-09-23T19:01:43.533290Z","iopub.status.idle":"2024-09-23T19:02:50.811241Z","shell.execute_reply.started":"2024-09-23T19:01:43.533250Z","shell.execute_reply":"2024-09-23T19:02:50.810539Z"},"trusted":true},"execution_count":21,"outputs":[]},{"cell_type":"code","source":"# apply tweet content cleaning functions to testing data\ntexts_new_test = []\nfor t in df_test.OriginalTweet:\n    texts_new_test.append(remove_mult_spaces(filter_chars(clean_hashtags(strip_all_entities(strip_emoji(t))))))","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:50.812341Z","iopub.execute_input":"2024-09-23T19:02:50.812569Z","iopub.status.idle":"2024-09-23T19:02:57.206976Z","shell.execute_reply.started":"2024-09-23T19:02:50.812543Z","shell.execute_reply":"2024-09-23T19:02:57.206334Z"},"trusted":true},"execution_count":22,"outputs":[]},{"cell_type":"code","source":"# new column (attribute) for cleaned tweets\ndf['text_clean'] = texts_new\ndf_test['text_clean'] = texts_new_test","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.208031Z","iopub.execute_input":"2024-09-23T19:02:57.208273Z","iopub.status.idle":"2024-09-23T19:02:57.219740Z","shell.execute_reply.started":"2024-09-23T19:02:57.208238Z","shell.execute_reply":"2024-09-23T19:02:57.218786Z"},"trusted":true},"execution_count":23,"outputs":[]},{"cell_type":"code","source":"df['text_clean'].head()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.221038Z","iopub.execute_input":"2024-09-23T19:02:57.221320Z","iopub.status.idle":"2024-09-23T19:02:57.232978Z","shell.execute_reply.started":"2024-09-23T19:02:57.221280Z","shell.execute_reply":"2024-09-23T19:02:57.232387Z"},"trusted":true},"execution_count":24,"outputs":[{"execution_count":24,"output_type":"execute_result","data":{"text/plain":"0                                              and and\n1    advice talk to your neighbours family to excha...\n2    coronavirus australia woolworths to give elder...\n3    my food stock is not the only one which is emp...\n4    me ready to go at supermarket during the covid...\nName: text_clean, dtype: object"},"metadata":{}}]},{"cell_type":"code","source":"df_test['text_clean'].head()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.233864Z","iopub.execute_input":"2024-09-23T19:02:57.234141Z","iopub.status.idle":"2024-09-23T19:02:57.242263Z","shell.execute_reply.started":"2024-09-23T19:02:57.234103Z","shell.execute_reply":"2024-09-23T19:02:57.241519Z"},"trusted":true},"execution_count":25,"outputs":[{"execution_count":25,"output_type":"execute_result","data":{"text/plain":"0    trending new yorkers encounter empty supermark...\n1    when i couldnt find hand sanitizer at fred mey...\n2    find out how you can protect yourself and love...\n3    panic buying hits newyork city as anxious shop...\n4    toiletpaper dunnypaper coronavirus coronavirus...\nName: text_clean, dtype: object"},"metadata":{}}]},{"cell_type":"code","source":"# compute tweet text length (to check if we removed too much of original tweet)\ntext_len = []\nfor text in df.text_clean:\n    tweet_len = len(text.split())\n    text_len.append(tweet_len)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.243347Z","iopub.execute_input":"2024-09-23T19:02:57.243663Z","iopub.status.idle":"2024-09-23T19:02:57.340717Z","shell.execute_reply.started":"2024-09-23T19:02:57.243622Z","shell.execute_reply":"2024-09-23T19:02:57.340132Z"},"trusted":true},"execution_count":26,"outputs":[]},{"cell_type":"code","source":"df['text_len'] = text_len","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.341775Z","iopub.execute_input":"2024-09-23T19:02:57.341997Z","iopub.status.idle":"2024-09-23T19:02:57.363624Z","shell.execute_reply.started":"2024-09-23T19:02:57.341972Z","shell.execute_reply":"2024-09-23T19:02:57.362998Z"},"trusted":true},"execution_count":27,"outputs":[]},{"cell_type":"code","source":"# compute tweet text length (to check if we removed too much of original tweet)\ntext_len_test = []\nfor text in df_test.text_clean:\n    tweet_len = len(text.split())\n    text_len_test.append(tweet_len)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.364548Z","iopub.execute_input":"2024-09-23T19:02:57.364727Z","iopub.status.idle":"2024-09-23T19:02:57.381570Z","shell.execute_reply.started":"2024-09-23T19:02:57.364703Z","shell.execute_reply":"2024-09-23T19:02:57.380957Z"},"trusted":true},"execution_count":28,"outputs":[]},{"cell_type":"code","source":"df_test['text_len'] = text_len_test","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.382441Z","iopub.execute_input":"2024-09-23T19:02:57.382673Z","iopub.status.idle":"2024-09-23T19:02:57.391645Z","shell.execute_reply.started":"2024-09-23T19:02:57.382647Z","shell.execute_reply":"2024-09-23T19:02:57.391001Z"},"trusted":true},"execution_count":29,"outputs":[]},{"cell_type":"code","source":"# plot count of training tweets with less than 10 words\nplt.figure(figsize=(7,5))\nax = sns.countplot(x='text_len', data=df[df['text_len']<10], palette='mako')\nplt.title('Training tweets with less than 10 words')\nplt.yticks([])\nax.bar_label(ax.containers[0])\nplt.ylabel('count')\nplt.xlabel('')\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:02:57.392554Z","iopub.execute_input":"2024-09-23T19:02:57.392756Z","iopub.status.idle":"2024-09-23T19:02:57.562680Z","shell.execute_reply.started":"2024-09-23T19:02:57.392731Z","shell.execute_reply":"2024-09-23T19:02:57.561980Z"},"trusted":true},"execution_count":30,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 504x360 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# plot count of testing tweets with less than 10 words\nplt.figure(figsize=(7,5))\nax = sns.countplot(x='text_len', data=df_test[df_test['text_len']<10], palette='mako')\nplt.title('Test tweets with less than 10 words')\nplt.yticks([])\nax.bar_label(ax.containers[0])\nplt.ylabel('count')\nplt.xlabel('')\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:02:57.563829Z","iopub.execute_input":"2024-09-23T19:02:57.564095Z","iopub.status.idle":"2024-09-23T19:02:57.717288Z","shell.execute_reply.started":"2024-09-23T19:02:57.564057Z","shell.execute_reply":"2024-09-23T19:02:57.716603Z"},"trusted":true},"execution_count":31,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 504x360 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# filter training tweets with length <= 4\ndf = df[df['text_len'] > 4]","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.718172Z","iopub.execute_input":"2024-09-23T19:02:57.718359Z","iopub.status.idle":"2024-09-23T19:02:57.726841Z","shell.execute_reply.started":"2024-09-23T19:02:57.718335Z","shell.execute_reply":"2024-09-23T19:02:57.726115Z"},"trusted":true},"execution_count":32,"outputs":[]},{"cell_type":"code","source":"# filter testing tweets with length <= 4\ndf_test = df_test[df_test['text_len'] > 4]","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:57.727869Z","iopub.execute_input":"2024-09-23T19:02:57.728086Z","iopub.status.idle":"2024-09-23T19:02:57.733222Z","shell.execute_reply.started":"2024-09-23T19:02:57.728060Z","shell.execute_reply":"2024-09-23T19:02:57.732523Z"},"trusted":true},"execution_count":33,"outputs":[]},{"cell_type":"code","source":"# tokenizer to split text into tokens that BERT can process\nfrom transformers import BertTokenizerFast\ntokenizer = BertTokenizerFast.from_pretrained('bert-base-uncased')","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2024-09-23T19:02:57.734245Z","iopub.execute_input":"2024-09-23T19:02:57.734457Z","iopub.status.idle":"2024-09-23T19:02:59.538086Z","shell.execute_reply.started":"2024-09-23T19:02:57.734431Z","shell.execute_reply":"2024-09-23T19:02:59.537509Z"},"trusted":true},"execution_count":34,"outputs":[{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/226k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"d01d9027bc7a4faeb33204e0a5deb214"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/455k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"e014059f58ba4b7999c3154241186dcc"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/48.0 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"87adc3220ba44966952cbf3deb5eb2fa"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/570 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"37eea32b60ee4734b5df7da7d2a19b29"}},"metadata":{}}]},{"cell_type":"code","source":"# take tweet texts in training data and encode them in tokenizer one at a time -> compute length and store it\n# find max token length\nimport numpy as np\ntoken_lens = []\n\nfor txt in df['text_clean'].values:\n    tokens = tokenizer.encode(txt, max_length=512, truncation=True)\n    token_lens.append(len(tokens))\n    \nmax_len=np.max(token_lens)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:02:59.539098Z","iopub.execute_input":"2024-09-23T19:02:59.539494Z","iopub.status.idle":"2024-09-23T19:03:07.244893Z","shell.execute_reply.started":"2024-09-23T19:02:59.539450Z","shell.execute_reply":"2024-09-23T19:03:07.244200Z"},"trusted":true},"execution_count":35,"outputs":[]},{"cell_type":"code","source":"# check training tweets tokenizes with more than 80 tokens\ntoken_lens = []\n\nfor i,txt in enumerate(df['text_clean'].values):\n    tokens = tokenizer.encode(txt, max_length=512, truncation=True)\n    token_lens.append(len(tokens))\n    if len(tokens)>80:\n        print(f\"INDEX: {i}, TEXT: {txt}\")  ","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:07.245931Z","iopub.execute_input":"2024-09-23T19:03:07.246163Z","iopub.status.idle":"2024-09-23T19:03:14.917162Z","shell.execute_reply.started":"2024-09-23T19:03:07.246134Z","shell.execute_reply":"2024-09-23T19:03:14.916524Z"},"trusted":true},"execution_count":36,"outputs":[{"name":"stdout","text":"INDEX: 1622, TEXT: zsah policie proti spekulantm s roukami na mj popud hejtman steckho kraje ve spoluprci s podle krizovho zkona zajistil tm 700 tisrouek od firmy kter je mla dodat na zdravotnkm ale na posledn chvli se snaila navyovat cenu spolutozvladneme\nINDEX: 13623, TEXT: hoy aplaudo a mi segunda familia aplaudoanuestrosheroes aquellos con los que he compartido tantas noches de trabajo y tanta alegra s que como siempre dan todo por el bien de su comunidad presidente por ellos tambin cuarentenanacionalya cuidemosalosquecuidan\nINDEX: 16548, TEXT: bir mddettir spermarketlerin lojistik hizmeti avusturya ordusu desteiyle yaplyor dn corona tedavisi iin 22 milyon luk bir aratrma gelitirme btesi aklad hkmet geen hafta da 35 milyon luk 2 yardm paketi aklanmt viyanadan haberler bu kadar\nINDEX: 36953, TEXT: 11 nisan cumartesi itibariyle bbnin tm hizmetleri sokaa kma serbestisi olanlar iin devam edecek halk ekmek ve hamidiye su 100 retime geti bb tm stanbulun gda ihtiyacna yetecek kapasitededir halkmz sakin olsun ve gvende hissetsin ltfen herkes evine dnsn\n","output_type":"stream"}]},{"cell_type":"markdown","source":"These sentences are not in english. They should be dropped.","metadata":{}},{"cell_type":"code","source":"df['token_lens'] = token_lens","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:14.918185Z","iopub.execute_input":"2024-09-23T19:03:14.918413Z","iopub.status.idle":"2024-09-23T19:03:14.940284Z","shell.execute_reply.started":"2024-09-23T19:03:14.918384Z","shell.execute_reply":"2024-09-23T19:03:14.939534Z"},"trusted":true},"execution_count":37,"outputs":[]},{"cell_type":"code","source":"df = df.sort_values(by='token_lens', ascending=False)\ndf.head(20)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:14.941218Z","iopub.execute_input":"2024-09-23T19:03:14.941407Z","iopub.status.idle":"2024-09-23T19:03:14.969193Z","shell.execute_reply.started":"2024-09-23T19:03:14.941383Z","shell.execute_reply":"2024-09-23T19:03:14.968581Z"},"trusted":true},"execution_count":38,"outputs":[{"execution_count":38,"output_type":"execute_result","data":{"text/plain":"                                           OriginalTweet           Sentiment  \\\n1638   ZÃ¡sah policie proti spekulant?m s rouÂkami. ...             Neutral   \n37156  11 Nisan Cumartesi itibariyle ?BBÂnin tÃ¼m hi...             Neutral   \n16632  Bir mÃ¼ddettir sÃ¼permarketlerin lojistik hizm...             Neutral   \n13691  Hoy aplaudo a mi segunda familia #AplaudoANues...             Neutral   \n27005  Supermercados Econo confirman que un empleado ...             Neutral   \n14593  Na, wer war denn da am Werk? Gestern Nachmitta...  Extremely Negative   \n28899  Kindly contact Us bamy global merchandise for ...            Positive   \n11213  Keine WertgegenstÃ¤nde im Fahrzeug lassen! - D...            Negative   \n4844   Impct of #coronavirus i hve sen hw civilizd pp...  Extremely Negative   \n18913  #CroozefmNews \\r\\r\\nPresident Museveni has ord...  Extremely Negative   \n30206  #LDA City Lahore Residential Files Prices Upda...             Neutral   \n26678  Eine wahre #CoronaGeschichte:\\r\\r\\n\\r\\r\\nWenn ...             Neutral   \n12389  Okay, so I just checked the drug prices for #P...            Positive   \n1697   I work at a grocery store.\\r\\r\\nWe wont get an...            Positive   \n8730   ?Bitte anschauen! (1/2)\\r\\r\\n\\r\\r\\nEmotionaler...            Negative   \n14582  hiked prices in the face of the Covid-19 crise...            Negative   \n36305  Sterile disposable anti bacterial wet wipes an...            Negative   \n36306  For sell Sterile disposable anti bacterial wet...            Negative   \n9238   Hi @Zomato I felt hungry loggd in ur app tryd ...  Extremely Positive   \n40778  #Covid_19 2days Stay-at-home activities\\r\\r\\nU...             Neutral   \n\n                                              text_clean  text_len  token_lens  \n1638   zsah policie proti spekulantm s roukami na mj ...        39         100  \n37156  11 nisan cumartesi itibariyle bbnin tm hizmetl...        39          98  \n16632  bir mddettir spermarketlerin lojistik hizmeti ...        36          92  \n13691  hoy aplaudo a mi segunda familia aplaudoanuest...        38          84  \n27005  supermercados econo confirman que un empleado ...        39          80  \n14593  na wer war denn da am werk gestern nachmittag ...        37          80  \n28899  kindly contact us bamy global merchandise for ...        37          80  \n11213  keine wertgegenstnde im fahrzeug lassen diesen...        33          79  \n4844   impct of coronavirus i hve sen hw civilizd ppl...        48          79  \n18913  croozefmnews president museveni has ordered th...        35          79  \n30206  lda city lahore residential files prices updat...        43          78  \n26678  eine wahre coronageschichte wenn dir an der su...        29          78  \n12389  okay so i just checked the drug prices for pla...        35          77  \n1697   i work at a grocery store we wont get any toil...        37          77  \n8730   bitte anschauen 12 emotionaler aufruf von geha...        36          77  \n14582  hiked prices in the face of the covid19 crises...        47          77  \n36305  sterile disposable anti bacterial wet wipes an...        31          76  \n36306  for sell sterile disposable anti bacterial wet...        32          75  \n9238   hi i felt hungry loggd in ur app tryd to add 4...        55          75  \n40778  covid19 2days stayathome activities up 630am f...        37          75  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>OriginalTweet</th>\n      <th>Sentiment</th>\n      <th>text_clean</th>\n      <th>text_len</th>\n      <th>token_lens</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>1638</th>\n      <td>ZÃ¡sah policie proti spekulant?m s rouÂkami. ...</td>\n      <td>Neutral</td>\n      <td>zsah policie proti spekulantm s roukami na mj ...</td>\n      <td>39</td>\n      <td>100</td>\n    </tr>\n    <tr>\n      <th>37156</th>\n      <td>11 Nisan Cumartesi itibariyle ?BBÂnin tÃ¼m hi...</td>\n      <td>Neutral</td>\n      <td>11 nisan cumartesi itibariyle bbnin tm hizmetl...</td>\n      <td>39</td>\n      <td>98</td>\n    </tr>\n    <tr>\n      <th>16632</th>\n      <td>Bir mÃ¼ddettir sÃ¼permarketlerin lojistik hizm...</td>\n      <td>Neutral</td>\n      <td>bir mddettir spermarketlerin lojistik hizmeti ...</td>\n      <td>36</td>\n      <td>92</td>\n    </tr>\n    <tr>\n      <th>13691</th>\n      <td>Hoy aplaudo a mi segunda familia #AplaudoANues...</td>\n      <td>Neutral</td>\n      <td>hoy aplaudo a mi segunda familia aplaudoanuest...</td>\n      <td>38</td>\n      <td>84</td>\n    </tr>\n    <tr>\n      <th>27005</th>\n      <td>Supermercados Econo confirman que un empleado ...</td>\n      <td>Neutral</td>\n      <td>supermercados econo confirman que un empleado ...</td>\n      <td>39</td>\n      <td>80</td>\n    </tr>\n    <tr>\n      <th>14593</th>\n      <td>Na, wer war denn da am Werk? Gestern Nachmitta...</td>\n      <td>Extremely Negative</td>\n      <td>na wer war denn da am werk gestern nachmittag ...</td>\n      <td>37</td>\n      <td>80</td>\n    </tr>\n    <tr>\n      <th>28899</th>\n      <td>Kindly contact Us bamy global merchandise for ...</td>\n      <td>Positive</td>\n      <td>kindly contact us bamy global merchandise for ...</td>\n      <td>37</td>\n      <td>80</td>\n    </tr>\n    <tr>\n      <th>11213</th>\n      <td>Keine WertgegenstÃ¤nde im Fahrzeug lassen! - D...</td>\n      <td>Negative</td>\n      <td>keine wertgegenstnde im fahrzeug lassen diesen...</td>\n      <td>33</td>\n      <td>79</td>\n    </tr>\n    <tr>\n      <th>4844</th>\n      <td>Impct of #coronavirus i hve sen hw civilizd pp...</td>\n      <td>Extremely Negative</td>\n      <td>impct of coronavirus i hve sen hw civilizd ppl...</td>\n      <td>48</td>\n      <td>79</td>\n    </tr>\n    <tr>\n      <th>18913</th>\n      <td>#CroozefmNews \\r\\r\\nPresident Museveni has ord...</td>\n      <td>Extremely Negative</td>\n      <td>croozefmnews president museveni has ordered th...</td>\n      <td>35</td>\n      <td>79</td>\n    </tr>\n    <tr>\n      <th>30206</th>\n      <td>#LDA City Lahore Residential Files Prices Upda...</td>\n      <td>Neutral</td>\n      <td>lda city lahore residential files prices updat...</td>\n      <td>43</td>\n      <td>78</td>\n    </tr>\n    <tr>\n      <th>26678</th>\n      <td>Eine wahre #CoronaGeschichte:\\r\\r\\n\\r\\r\\nWenn ...</td>\n      <td>Neutral</td>\n      <td>eine wahre coronageschichte wenn dir an der su...</td>\n      <td>29</td>\n      <td>78</td>\n    </tr>\n    <tr>\n      <th>12389</th>\n      <td>Okay, so I just checked the drug prices for #P...</td>\n      <td>Positive</td>\n      <td>okay so i just checked the drug prices for pla...</td>\n      <td>35</td>\n      <td>77</td>\n    </tr>\n    <tr>\n      <th>1697</th>\n      <td>I work at a grocery store.\\r\\r\\nWe wont get an...</td>\n      <td>Positive</td>\n      <td>i work at a grocery store we wont get any toil...</td>\n      <td>37</td>\n      <td>77</td>\n    </tr>\n    <tr>\n      <th>8730</th>\n      <td>?Bitte anschauen! (1/2)\\r\\r\\n\\r\\r\\nEmotionaler...</td>\n      <td>Negative</td>\n      <td>bitte anschauen 12 emotionaler aufruf von geha...</td>\n      <td>36</td>\n      <td>77</td>\n    </tr>\n    <tr>\n      <th>14582</th>\n      <td>hiked prices in the face of the Covid-19 crise...</td>\n      <td>Negative</td>\n      <td>hiked prices in the face of the covid19 crises...</td>\n      <td>47</td>\n      <td>77</td>\n    </tr>\n    <tr>\n      <th>36305</th>\n      <td>Sterile disposable anti bacterial wet wipes an...</td>\n      <td>Negative</td>\n      <td>sterile disposable anti bacterial wet wipes an...</td>\n      <td>31</td>\n      <td>76</td>\n    </tr>\n    <tr>\n      <th>36306</th>\n      <td>For sell Sterile disposable anti bacterial wet...</td>\n      <td>Negative</td>\n      <td>for sell sterile disposable anti bacterial wet...</td>\n      <td>32</td>\n      <td>75</td>\n    </tr>\n    <tr>\n      <th>9238</th>\n      <td>Hi @Zomato I felt hungry loggd in ur app tryd ...</td>\n      <td>Extremely Positive</td>\n      <td>hi i felt hungry loggd in ur app tryd to add 4...</td>\n      <td>55</td>\n      <td>75</td>\n    </tr>\n    <tr>\n      <th>40778</th>\n      <td>#Covid_19 2days Stay-at-home activities\\r\\r\\nU...</td>\n      <td>Neutral</td>\n      <td>covid19 2days stayathome activities up 630am f...</td>\n      <td>37</td>\n      <td>75</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"df = df.sample(frac=1).reset_index(drop=True)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:14.970202Z","iopub.execute_input":"2024-09-23T19:03:14.970410Z","iopub.status.idle":"2024-09-23T19:03:14.983391Z","shell.execute_reply.started":"2024-09-23T19:03:14.970384Z","shell.execute_reply":"2024-09-23T19:03:14.982646Z"},"trusted":true},"execution_count":39,"outputs":[]},{"cell_type":"code","source":"# take tweet texts in testing data and encode them in tokenizer one at a time -> compute length and store it\n# find max token length\ntoken_lens_test = []\n\nfor txt in df_test['text_clean'].values:\n    tokens = tokenizer.encode(txt, max_length=512, truncation=True)\n    token_lens_test.append(len(tokens))\n    \nmax_len=np.max(token_lens_test)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:14.984285Z","iopub.execute_input":"2024-09-23T19:03:14.984515Z","iopub.status.idle":"2024-09-23T19:03:15.730912Z","shell.execute_reply.started":"2024-09-23T19:03:14.984463Z","shell.execute_reply":"2024-09-23T19:03:15.730291Z"},"trusted":true},"execution_count":40,"outputs":[]},{"cell_type":"code","source":"# check testing tweets tokenizes with more than 80 tokens\ntoken_lens_test = []\n\nfor i,txt in enumerate(df_test['text_clean'].values):\n    tokens = tokenizer.encode(txt, max_length=512, truncation=True)\n    token_lens_test.append(len(tokens))\n    if len(tokens)>80:\n        print(f\"INDEX: {i}, TEXT: {txt}\")","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:15.732113Z","iopub.execute_input":"2024-09-23T19:03:15.732887Z","iopub.status.idle":"2024-09-23T19:03:16.478849Z","shell.execute_reply.started":"2024-09-23T19:03:15.732836Z","shell.execute_reply":"2024-09-23T19:03:16.477984Z"},"trusted":true},"execution_count":41,"outputs":[{"name":"stdout","text":"INDEX: 286, TEXT: so hard to decide as much as i want to hodl my 2 ccdcv4 token our place is declared to lock down due to covid19 i will use this to buy some food to stock txnid093bd1db0c0d3a62af15883138a5f57d4cef35ae14e31e602b74489dd2524c7f my b\nINDEX: 345, TEXT: informoval jsem zstupce vech obchodnch etzc o aktulnch opatench vldy etzce jsou zsobovny na 95 take nen dvod panikait zsoby potravin fakt nedojdou nen opravdu dvod dnes obsadit a vykoupit supermarkety\nINDEX: 2380, TEXT: ahora seguid llorando por el papel higinico que no he comprado porque an tengo seguid creando histeria y preocupacin poniendo fotos de gente en pnico y estanteras vacas que yo seguir yendo a comercios responsables de barrio donde nos cuidan hoy y siempre gracias\n","output_type":"stream"}]},{"cell_type":"code","source":"df_test['token_lens'] = token_lens_test","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.479938Z","iopub.execute_input":"2024-09-23T19:03:16.480134Z","iopub.status.idle":"2024-09-23T19:03:16.486737Z","shell.execute_reply.started":"2024-09-23T19:03:16.480109Z","shell.execute_reply":"2024-09-23T19:03:16.485934Z"},"trusted":true},"execution_count":42,"outputs":[]},{"cell_type":"code","source":"df_test = df_test.sort_values(by='token_lens', ascending=False)\ndf_test.head(10) ","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.487686Z","iopub.execute_input":"2024-09-23T19:03:16.487932Z","iopub.status.idle":"2024-09-23T19:03:16.504655Z","shell.execute_reply.started":"2024-09-23T19:03:16.487899Z","shell.execute_reply":"2024-09-23T19:03:16.503961Z"},"trusted":true},"execution_count":43,"outputs":[{"execution_count":43,"output_type":"execute_result","data":{"text/plain":"                                          OriginalTweet           Sentiment  \\\n286   @Rhett800cc So hard to decide??. As much as I ...            Negative   \n2383  Ahora seguid llorando por el papel higiénico (...            Negative   \n345   Informoval jsem zástupce vech obchodních ?et?...             Neutral   \n1485  DTF-Dont Touch Face\\r\\r\\nDWBH-Do Wash Both Ha...  Extremely Negative   \n1209  Im in the DC/Maryland/Virginia (DMV) area &am...            Positive   \n3505  Stop misusing ur privilege amp grow up Some1 c...            Positive   \n1789  For those that are cashlong, patient,calm&amp;...  Extremely Positive   \n855   Lidl is total chaos, queues as long as the ais...  Extremely Negative   \n2740  COVID-19: Your government will save ITSELF not...            Positive   \n2997  Stop #frenzybuying. You don't need most of wha...  Extremely Negative   \n\n                                             text_clean  text_len  token_lens  \n286   so hard to decide as much as i want to hodl my...        38          96  \n2383  ahora seguid llorando por el papel higinico qu...        44          94  \n345   informoval jsem zstupce vech obchodnch etzc o ...        31          86  \n1485  dtfdont touch face dwbhdo wash both hands gtfo...        42          77  \n1209  im in the dcmarylandvirginia dmv area amphave ...        45          74  \n3505  stop misusing ur privilege amp grow up some1 c...        57          73  \n1789  for those that are cashlong patientcalmamphave...        44          71  \n855   lidl is total chaos queues as long as the aisl...        62          70  \n2740  covid19 your government will save itself not y...        43          70  \n2997  stop frenzybuying you dont need most of what y...        38          70  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>OriginalTweet</th>\n      <th>Sentiment</th>\n      <th>text_clean</th>\n      <th>text_len</th>\n      <th>token_lens</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>286</th>\n      <td>@Rhett800cc So hard to decide??. As much as I ...</td>\n      <td>Negative</td>\n      <td>so hard to decide as much as i want to hodl my...</td>\n      <td>38</td>\n      <td>96</td>\n    </tr>\n    <tr>\n      <th>2383</th>\n      <td>Ahora seguid llorando por el papel higiénico (...</td>\n      <td>Negative</td>\n      <td>ahora seguid llorando por el papel higinico qu...</td>\n      <td>44</td>\n      <td>94</td>\n    </tr>\n    <tr>\n      <th>345</th>\n      <td>Informoval jsem zástupce vech obchodních ?et?...</td>\n      <td>Neutral</td>\n      <td>informoval jsem zstupce vech obchodnch etzc o ...</td>\n      <td>31</td>\n      <td>86</td>\n    </tr>\n    <tr>\n      <th>1485</th>\n      <td>DTF-Dont Touch Face\\r\\r\\nDWBH-Do Wash Both Ha...</td>\n      <td>Extremely Negative</td>\n      <td>dtfdont touch face dwbhdo wash both hands gtfo...</td>\n      <td>42</td>\n      <td>77</td>\n    </tr>\n    <tr>\n      <th>1209</th>\n      <td>Im in the DC/Maryland/Virginia (DMV) area &amp;am...</td>\n      <td>Positive</td>\n      <td>im in the dcmarylandvirginia dmv area amphave ...</td>\n      <td>45</td>\n      <td>74</td>\n    </tr>\n    <tr>\n      <th>3505</th>\n      <td>Stop misusing ur privilege amp grow up Some1 c...</td>\n      <td>Positive</td>\n      <td>stop misusing ur privilege amp grow up some1 c...</td>\n      <td>57</td>\n      <td>73</td>\n    </tr>\n    <tr>\n      <th>1789</th>\n      <td>For those that are cashlong, patient,calm&amp;amp;...</td>\n      <td>Extremely Positive</td>\n      <td>for those that are cashlong patientcalmamphave...</td>\n      <td>44</td>\n      <td>71</td>\n    </tr>\n    <tr>\n      <th>855</th>\n      <td>Lidl is total chaos, queues as long as the ais...</td>\n      <td>Extremely Negative</td>\n      <td>lidl is total chaos queues as long as the aisl...</td>\n      <td>62</td>\n      <td>70</td>\n    </tr>\n    <tr>\n      <th>2740</th>\n      <td>COVID-19: Your government will save ITSELF not...</td>\n      <td>Positive</td>\n      <td>covid19 your government will save itself not y...</td>\n      <td>43</td>\n      <td>70</td>\n    </tr>\n    <tr>\n      <th>2997</th>\n      <td>Stop #frenzybuying. You don't need most of wha...</td>\n      <td>Extremely Negative</td>\n      <td>stop frenzybuying you dont need most of what y...</td>\n      <td>38</td>\n      <td>70</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"df_test = df_test.sample(frac=1).reset_index(drop=True)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.505720Z","iopub.execute_input":"2024-09-23T19:03:16.505983Z","iopub.status.idle":"2024-09-23T19:03:16.511970Z","shell.execute_reply.started":"2024-09-23T19:03:16.505945Z","shell.execute_reply":"2024-09-23T19:03:16.511278Z"},"trusted":true},"execution_count":44,"outputs":[]},{"cell_type":"code","source":"# count number of positive, negative, neutral, etc tweets\ndf['Sentiment'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.519565Z","iopub.execute_input":"2024-09-23T19:03:16.519782Z","iopub.status.idle":"2024-09-23T19:03:16.530580Z","shell.execute_reply.started":"2024-09-23T19:03:16.519756Z","shell.execute_reply":"2024-09-23T19:03:16.529828Z"},"trusted":true},"execution_count":45,"outputs":[{"execution_count":45,"output_type":"execute_result","data":{"text/plain":"Positive              11382\nNegative               9890\nNeutral                7567\nExtremely Positive     6618\nExtremely Negative     5478\nName: Sentiment, dtype: int64"},"metadata":{}}]},{"cell_type":"markdown","source":"The first thing we can do is to encode the categories with numbers. We will also create just 3 possible emotions: Positive, Neutral and Negative.","metadata":{"execution":{"iopub.status.busy":"2021-12-05T11:35:57.747932Z","iopub.execute_input":"2021-12-05T11:35:57.748569Z","iopub.status.idle":"2021-12-05T11:35:57.753489Z","shell.execute_reply.started":"2021-12-05T11:35:57.748528Z","shell.execute_reply":"2021-12-05T11:35:57.752685Z"}}},{"cell_type":"code","source":"# encode categories with numbers (just three categories) on training data\ndf['Sentiment'] = df['Sentiment'].map({'Extremely Negative':0,'Negative':0,'Neutral':1,'Positive':2,'Extremely Positive':2})","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.531557Z","iopub.execute_input":"2024-09-23T19:03:16.531802Z","iopub.status.idle":"2024-09-23T19:03:16.546112Z","shell.execute_reply.started":"2024-09-23T19:03:16.531765Z","shell.execute_reply":"2024-09-23T19:03:16.545396Z"},"trusted":true},"execution_count":46,"outputs":[]},{"cell_type":"code","source":"# encode categories with numbers (just three categories) on testing data\ndf_test['Sentiment'] = df_test['Sentiment'].map({'Extremely Negative':0,'Negative':0,'Neutral':1,'Positive':2,'Extremely Positive':2})","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.547356Z","iopub.execute_input":"2024-09-23T19:03:16.547735Z","iopub.status.idle":"2024-09-23T19:03:16.553590Z","shell.execute_reply.started":"2024-09-23T19:03:16.547691Z","shell.execute_reply":"2024-09-23T19:03:16.552884Z"},"trusted":true},"execution_count":47,"outputs":[]},{"cell_type":"code","source":"df['Sentiment'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.554590Z","iopub.execute_input":"2024-09-23T19:03:16.554795Z","iopub.status.idle":"2024-09-23T19:03:16.564722Z","shell.execute_reply.started":"2024-09-23T19:03:16.554769Z","shell.execute_reply":"2024-09-23T19:03:16.564006Z"},"trusted":true},"execution_count":48,"outputs":[{"execution_count":48,"output_type":"execute_result","data":{"text/plain":"2    18000\n0    15368\n1     7567\nName: Sentiment, dtype: int64"},"metadata":{}}]},{"cell_type":"code","source":"# since the classes are imbalanced -> oversample train test to remove bias towards majority class\nfrom imblearn.over_sampling import RandomOverSampler\nros = RandomOverSampler()\ntrain_x, train_y = ros.fit_resample(np.array(df['text_clean']).reshape(-1, 1), np.array(df['Sentiment']).reshape(-1, 1));\ntrain_os = pd.DataFrame(list(zip([x[0] for x in train_x], train_y)), columns = ['text_clean', 'Sentiment']);","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.565507Z","iopub.execute_input":"2024-09-23T19:03:16.565717Z","iopub.status.idle":"2024-09-23T19:03:16.868501Z","shell.execute_reply.started":"2024-09-23T19:03:16.565691Z","shell.execute_reply":"2024-09-23T19:03:16.867802Z"},"trusted":true},"execution_count":49,"outputs":[]},{"cell_type":"code","source":"train_os['Sentiment'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.869559Z","iopub.execute_input":"2024-09-23T19:03:16.869786Z","iopub.status.idle":"2024-09-23T19:03:16.876805Z","shell.execute_reply.started":"2024-09-23T19:03:16.869759Z","shell.execute_reply":"2024-09-23T19:03:16.876116Z"},"trusted":true},"execution_count":50,"outputs":[{"execution_count":50,"output_type":"execute_result","data":{"text/plain":"2    18000\n0    18000\n1    18000\nName: Sentiment, dtype: int64"},"metadata":{}}]},{"cell_type":"code","source":"X = train_os['text_clean'].values\ny = train_os['Sentiment'].values","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.877876Z","iopub.execute_input":"2024-09-23T19:03:16.878099Z","iopub.status.idle":"2024-09-23T19:03:16.884789Z","shell.execute_reply.started":"2024-09-23T19:03:16.878072Z","shell.execute_reply":"2024-09-23T19:03:16.884043Z"},"trusted":true},"execution_count":51,"outputs":[]},{"cell_type":"code","source":"# validation set extracted from training set to prevent overfitting\nfrom sklearn.model_selection import train_test_split\nseed=43\nX_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.1, stratify=y, random_state=seed)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.885821Z","iopub.execute_input":"2024-09-23T19:03:16.886076Z","iopub.status.idle":"2024-09-23T19:03:16.932588Z","shell.execute_reply.started":"2024-09-23T19:03:16.886039Z","shell.execute_reply":"2024-09-23T19:03:16.931889Z"},"trusted":true},"execution_count":52,"outputs":[]},{"cell_type":"code","source":"X_test = df_test['text_clean'].values\ny_test = df_test['Sentiment'].values","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.933570Z","iopub.execute_input":"2024-09-23T19:03:16.933817Z","iopub.status.idle":"2024-09-23T19:03:16.937618Z","shell.execute_reply.started":"2024-09-23T19:03:16.933784Z","shell.execute_reply":"2024-09-23T19:03:16.936932Z"},"trusted":true},"execution_count":53,"outputs":[]},{"cell_type":"code","source":"# save a copy of label encoded Sentiment\ny_train_le = y_train.copy()\ny_valid_le = y_valid.copy()\ny_test_le = y_test.copy()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.938717Z","iopub.execute_input":"2024-09-23T19:03:16.938988Z","iopub.status.idle":"2024-09-23T19:03:16.946086Z","shell.execute_reply.started":"2024-09-23T19:03:16.938953Z","shell.execute_reply":"2024-09-23T19:03:16.945366Z"},"trusted":true},"execution_count":54,"outputs":[]},{"cell_type":"code","source":"# use one-hot encoding for Sentiment (same as dummies)\nfrom sklearn import preprocessing\nohe = preprocessing.OneHotEncoder()\ny_train = ohe.fit_transform(np.array(y_train).reshape(-1, 1)).toarray()\ny_valid = ohe.fit_transform(np.array(y_valid).reshape(-1, 1)).toarray()\ny_test = ohe.fit_transform(np.array(y_test).reshape(-1, 1)).toarray()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.947000Z","iopub.execute_input":"2024-09-23T19:03:16.947199Z","iopub.status.idle":"2024-09-23T19:03:16.961123Z","shell.execute_reply.started":"2024-09-23T19:03:16.947175Z","shell.execute_reply":"2024-09-23T19:03:16.960353Z"},"trusted":true},"execution_count":55,"outputs":[]},{"cell_type":"markdown","source":"# MODEL 0: NAIVE BAYES CLASSIFIER","metadata":{}},{"cell_type":"code","source":"# tokenize tweets\nfrom sklearn.feature_extraction.text import CountVectorizer\nclf = CountVectorizer()\nX_train_cv =  clf.fit_transform(X_train)\nX_test_cv = clf.transform(X_test)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:16.961998Z","iopub.execute_input":"2024-09-23T19:03:16.962200Z","iopub.status.idle":"2024-09-23T19:03:18.952434Z","shell.execute_reply.started":"2024-09-23T19:03:16.962175Z","shell.execute_reply":"2024-09-23T19:03:18.951804Z"},"trusted":true},"execution_count":56,"outputs":[]},{"cell_type":"code","source":"# TF-IDF to evaluate importance of a word\nfrom sklearn.feature_extraction.text import TfidfTransformer\ntf_transformer = TfidfTransformer(use_idf=True).fit(X_train_cv)\nX_train_tf = tf_transformer.transform(X_train_cv)\nX_test_tf = tf_transformer.transform(X_test_cv)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:18.953584Z","iopub.execute_input":"2024-09-23T19:03:18.953852Z","iopub.status.idle":"2024-09-23T19:03:19.070930Z","shell.execute_reply.started":"2024-09-23T19:03:18.953816Z","shell.execute_reply":"2024-09-23T19:03:19.070284Z"},"trusted":true},"execution_count":57,"outputs":[]},{"cell_type":"code","source":"# define Naive Bayes Classifier model\nfrom sklearn.naive_bayes import MultinomialNB\nnb_clf = MultinomialNB()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:19.072047Z","iopub.execute_input":"2024-09-23T19:03:19.072366Z","iopub.status.idle":"2024-09-23T19:03:19.078857Z","shell.execute_reply.started":"2024-09-23T19:03:19.072328Z","shell.execute_reply":"2024-09-23T19:03:19.078067Z"},"trusted":true},"execution_count":58,"outputs":[]},{"cell_type":"code","source":"nb_clf.fit(X_train_tf, y_train_le)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:19.079842Z","iopub.execute_input":"2024-09-23T19:03:19.080052Z","iopub.status.idle":"2024-09-23T19:03:19.108730Z","shell.execute_reply.started":"2024-09-23T19:03:19.080026Z","shell.execute_reply":"2024-09-23T19:03:19.108026Z"},"trusted":true},"execution_count":59,"outputs":[{"execution_count":59,"output_type":"execute_result","data":{"text/plain":"MultinomialNB()"},"metadata":{}}]},{"cell_type":"code","source":"nb_pred = nb_clf.predict(X_test_tf)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:19.109585Z","iopub.execute_input":"2024-09-23T19:03:19.109767Z","iopub.status.idle":"2024-09-23T19:03:19.115565Z","shell.execute_reply.started":"2024-09-23T19:03:19.109744Z","shell.execute_reply":"2024-09-23T19:03:19.114800Z"},"trusted":true},"execution_count":60,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import classification_report, accuracy_score, f1_score\nprint('\\tClassification Report for Naive Bayes\\n\\n',classification_report(y_test_le,nb_pred, target_names=['Negative', 'Neutral', 'Positive']))","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:03:19.116812Z","iopub.execute_input":"2024-09-23T19:03:19.117042Z","iopub.status.idle":"2024-09-23T19:03:19.130824Z","shell.execute_reply.started":"2024-09-23T19:03:19.117015Z","shell.execute_reply":"2024-09-23T19:03:19.130056Z"},"trusted":true},"execution_count":61,"outputs":[{"name":"stdout","text":"\tClassification Report for Naive Bayes\n\n               precision    recall  f1-score   support\n\n    Negative       0.71      0.78      0.74      1632\n     Neutral       0.60      0.45      0.51       615\n    Positive       0.73      0.73      0.73      1545\n\n    accuracy                           0.71      3792\n   macro avg       0.68      0.65      0.66      3792\nweighted avg       0.70      0.71      0.70      3792\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# MODEL 1: BERT","metadata":{}},{"cell_type":"code","source":"# custom tokenize function to generate tokens that BERT can process\nMAX_LEN=128 # max length of tokenizes sequences (max number of tokens given to BERT)\ndef tokenize(data,max_len=MAX_LEN) :\n    input_ids = []\n    attention_masks = []\n    for i in range(len(data)):\n        encoded = tokenizer.encode_plus(\n            data[i],\n            add_special_tokens=True,\n            max_length=MAX_LEN,\n            padding='max_length',\n            return_attention_mask=True\n        )\n        input_ids.append(encoded['input_ids'])\n        attention_masks.append(encoded['attention_mask'])\n    return np.array(input_ids),np.array(attention_masks)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:19.131753Z","iopub.execute_input":"2024-09-23T19:03:19.131964Z","iopub.status.idle":"2024-09-23T19:03:19.139239Z","shell.execute_reply.started":"2024-09-23T19:03:19.131938Z","shell.execute_reply":"2024-09-23T19:03:19.138576Z"},"trusted":true},"execution_count":62,"outputs":[]},{"cell_type":"code","source":"# tokenize train, validation, test datasets\ntrain_input_ids, train_attention_masks = tokenize(X_train, MAX_LEN)\nval_input_ids, val_attention_masks = tokenize(X_valid, MAX_LEN)\ntest_input_ids, test_attention_masks = tokenize(X_test, MAX_LEN)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:19.140267Z","iopub.execute_input":"2024-09-23T19:03:19.140768Z","iopub.status.idle":"2024-09-23T19:03:34.780059Z","shell.execute_reply.started":"2024-09-23T19:03:19.140730Z","shell.execute_reply":"2024-09-23T19:03:34.779413Z"},"trusted":true},"execution_count":63,"outputs":[]},{"cell_type":"code","source":"# define BERT model\nfrom transformers import TFBertModel\nbert_model = TFBertModel.from_pretrained('bert-base-uncased')","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2024-09-23T19:03:34.781135Z","iopub.execute_input":"2024-09-23T19:03:34.781381Z","iopub.status.idle":"2024-09-23T19:03:56.083122Z","shell.execute_reply.started":"2024-09-23T19:03:34.781351Z","shell.execute_reply":"2024-09-23T19:03:56.082406Z"},"trusted":true},"execution_count":64,"outputs":[{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/511M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"2646993ef9c44e1f976cddf99b50b446"}},"metadata":{}},{"name":"stderr","text":"Some layers from the model checkpoint at bert-base-uncased were not used when initializing TFBertModel: ['nsp___cls', 'mlm___cls']\n- This IS expected if you are initializing TFBertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n- This IS NOT expected if you are initializing TFBertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\nAll the layers of TFBertModel were initialized from the model checkpoint at bert-base-uncased.\nIf your task is similar to the task the model of the checkpoint was trained on, you can already use TFBertModel for predictions without further training.\n","output_type":"stream"}]},{"cell_type":"code","source":"# function to host pre-trained BERT model and attach it to a 3 neurons output layer,\n# necessary to perform the classification of the 3 classes (emotions) of the dataset\nimport tensorflow as tf\nfrom tensorflow import keras\ndef create_model(bert_model, max_len=MAX_LEN):\n    \n    ##params###\n    opt = tf.keras.optimizers.Adam(learning_rate=1e-5, decay=1e-7)\n    loss = tf.keras.losses.CategoricalCrossentropy()\n    accuracy = tf.keras.metrics.CategoricalAccuracy()\n\n\n    input_ids = tf.keras.Input(shape=(max_len,),dtype='int32')\n    \n    attention_masks = tf.keras.Input(shape=(max_len,),dtype='int32')\n    \n    embeddings = bert_model([input_ids,attention_masks])[1]\n    \n    output = tf.keras.layers.Dense(3, activation=\"softmax\")(embeddings)\n    \n    model = tf.keras.models.Model(inputs = [input_ids,attention_masks], outputs = output)\n    \n    model.compile(opt, loss=loss, metrics=accuracy)\n    \n    \n    return model","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:56.084134Z","iopub.execute_input":"2024-09-23T19:03:56.084343Z","iopub.status.idle":"2024-09-23T19:03:56.092459Z","shell.execute_reply.started":"2024-09-23T19:03:56.084317Z","shell.execute_reply":"2024-09-23T19:03:56.091753Z"},"trusted":true},"execution_count":65,"outputs":[]},{"cell_type":"code","source":"model = create_model(bert_model, MAX_LEN)\nmodel.summary()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:03:56.093709Z","iopub.execute_input":"2024-09-23T19:03:56.094023Z","iopub.status.idle":"2024-09-23T19:04:03.512129Z","shell.execute_reply.started":"2024-09-23T19:03:56.093987Z","shell.execute_reply":"2024-09-23T19:04:03.511339Z"},"trusted":true},"execution_count":66,"outputs":[{"name":"stdout","text":"Model: \"model\"\n__________________________________________________________________________________________________\nLayer (type)                    Output Shape         Param #     Connected to                     \n==================================================================================================\ninput_1 (InputLayer)            [(None, 128)]        0                                            \n__________________________________________________________________________________________________\ninput_2 (InputLayer)            [(None, 128)]        0                                            \n__________________________________________________________________________________________________\ntf_bert_model (TFBertModel)     TFBaseModelOutputWit 109482240   input_1[0][0]                    \n                                                                 input_2[0][0]                    \n__________________________________________________________________________________________________\ndense (Dense)                   (None, 3)            2307        tf_bert_model[0][1]              \n==================================================================================================\nTotal params: 109,484,547\nTrainable params: 109,484,547\nNon-trainable params: 0\n__________________________________________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"history_bert = model.fit([train_input_ids,train_attention_masks], y_train, validation_data=([val_input_ids,val_attention_masks], y_valid), epochs=4, batch_size=32)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:04:03.513348Z","iopub.execute_input":"2024-09-23T19:04:03.513657Z","iopub.status.idle":"2024-09-23T19:53:22.822632Z","shell.execute_reply.started":"2024-09-23T19:04:03.513617Z","shell.execute_reply":"2024-09-23T19:53:22.821731Z"},"trusted":true},"execution_count":67,"outputs":[{"name":"stdout","text":"Epoch 1/4\n1519/1519 [==============================] - 753s 486ms/step - loss: 0.5500 - categorical_accuracy: 0.7825 - val_loss: 0.3503 - val_categorical_accuracy: 0.8735\nEpoch 2/4\n1519/1519 [==============================] - 735s 484ms/step - loss: 0.2921 - categorical_accuracy: 0.8979 - val_loss: 0.2484 - val_categorical_accuracy: 0.9069\nEpoch 3/4\n1519/1519 [==============================] - 735s 484ms/step - loss: 0.1971 - categorical_accuracy: 0.9313 - val_loss: 0.2198 - val_categorical_accuracy: 0.9217\nEpoch 4/4\n1519/1519 [==============================] - 736s 484ms/step - loss: 0.1339 - categorical_accuracy: 0.9542 - val_loss: 0.2257 - val_categorical_accuracy: 0.9302\n","output_type":"stream"}]},{"cell_type":"code","source":"result_bert = model.predict([test_input_ids,test_attention_masks])","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:53:22.824395Z","iopub.execute_input":"2024-09-23T19:53:22.824810Z","iopub.status.idle":"2024-09-23T19:53:44.326364Z","shell.execute_reply.started":"2024-09-23T19:53:22.824770Z","shell.execute_reply":"2024-09-23T19:53:44.325535Z"},"trusted":true},"execution_count":68,"outputs":[]},{"cell_type":"code","source":"# convert model's predicted probability distributions (result_bert)\n# into one-hot encoded labels, where each row represents a single sample\n# and contains a 1 in the column of the predicted class and 0s elsewhere\ny_pred_bert =  np.zeros_like(result_bert)\ny_pred_bert[np.arange(len(y_pred_bert)), result_bert.argmax(1)] = 1","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:53:44.327951Z","iopub.execute_input":"2024-09-23T19:53:44.328574Z","iopub.status.idle":"2024-09-23T19:53:44.333322Z","shell.execute_reply.started":"2024-09-23T19:53:44.328531Z","shell.execute_reply":"2024-09-23T19:53:44.332598Z"},"trusted":true},"execution_count":69,"outputs":[]},{"cell_type":"code","source":"# confusion matrix\nfrom sklearn.metrics import confusion_matrix\nconf_matrix(y_test.argmax(1), y_pred_bert.argmax(1),'BERT Sentiment Analysis\\nConfusion Matrix')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:53:44.334387Z","iopub.execute_input":"2024-09-23T19:53:44.334668Z","iopub.status.idle":"2024-09-23T19:53:44.470357Z","shell.execute_reply.started":"2024-09-23T19:53:44.334632Z","shell.execute_reply":"2024-09-23T19:53:44.469685Z"},"trusted":true},"execution_count":70,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 360x360 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAVUAAAF5CAYAAAAvYrljAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAABdvUlEQVR4nO2dd3gU1deA35MECB3pTTrSVASxgKigWEAUFBV77733n4Ig9u5nF7BhAxRURAQBCyBSVYpKk957CqSc7487Gzab3c0mbLKbcN7nmWd35pY5Mzt75pZzzxFVxTAMw4gOCbEWwDAMozRhStUwDCOKmFI1DMOIIqZUDcMwoogpVcMwjChiStUwDCOKmFI1SiwiskJEVsRaDiMvIjJcRFREmhTT+eLmWTClmg/egxG47fF+xPdFpE2QMt1ClMu1BZQZHiRPqogsFJHnRaRWQeoO2JpEcJ2tROQdEVkiIukikiIiy0Vkgog8KiJ1onZTI0REpgTep9JItBSQiPzg1bNKRBKjJJ5RQJJiLUAJYqDf96rA0cBlQD8R6aqq84KU+Q8YXsDzjAF8ddUBegF3eec5ElgRIAtANeB2YAfwUpA6t4c7oYicBHwLJAPTgfHATqA+0AU4BZgGbCjQlRQ9J8dagHhBRJrh7ocCDYGewDcxFap4iZ9nQVVtC7PhHlINkfaqlz484Hg37/iUApxnuFfmioDjyTglq8BjIco28dJXFPIa//XKXx4i/XDg4Bjc+ymh7n1p2vx++yb7UceTXh2+z7El/ZpK6mbd//1jgvdZq6hOoKrpwMfe7lHRrl9EagMtgB2q+n4IGf5Q1VVByjYUkddEZJk3JLJFRMaKSB45RWSA1zXtJiLnishMb3hjq4h8KiIN/PI28br9J3r7/kMZU/zy5RlHE5ErvHxXiMgpIvKziOwWkU0iMkxEqnn5OojINyKyzUsfG6r7LSLVReRJEVkkImkiskNEJonIqUHy+p+/uzeEsUtEdorIt4HDRd51Xu7tLve7zhWBdYdCRJKAK3C9i8eB2UAv/3sakH+Kd44kEXlIRP71fr9VIvK0iJQNUqaviHwkIv94Q0MpIjJbRG4TkXz1iIi09s45OUyeP0UkQ0TqefsiIpeLyDTv90v3ZPxeRPoHlA32LJT15Jvj/c6pXr4xItIjP5kLi3X/9w/fDzOrmM6XUQR17gAygUoiUk9V10VSSEQ64l4q1YHvgdFATaAv8IuInK2q44IUvQk4CxgLTAWOAfoD7UXkCFXdgxuuGIhTFI3JPdyxIsLrOgvojesCv4kbxrgCaCIiDwKTgJ+B94DDgDOBZiJyuKpm+11nY1yLuYmXfzxQ0at7vIhcr6rvBDl/b6AP8J13/ra4oZyjRKStqm728g3E3bP2wMvsG6rZTuScBdQF3lHVNBEZjutFXQUMClNuBHC8J+NOT777gNrAlQF5nwKygd+ANbghsJM8mY8CLg0noKou9hRqdxE5RFX/8U8XkS7AocAov2fwCeBBYDnwOe5Zreed7zzgs3DnxLWWLwT+Aj4A0nBDWl2B04GJ+ZQvHLFuKsf7htf9Bwb4bS/g/mDZwNdA5YAy3bwyKwLK+W8XBJQZTvDuf3ngDy/t7hAyNmH/uv8jvfJLgXtwiq5CmPxJwBIgHTgxIK0+7k+3Dijnd3yAd46dwGEBZUZ4aecHHJ9CmO6/d39XBBy7wqsr01823KTsD17aVuDigHLveWl9gsiQHeT3qoYblkkD6oQ4/8kBZXxd8/tC/PZNCvn7jffKd/b2qwN7vPuTECT/FC//bKC63/GK3u+aBdQNKNM8SD0JwPteXcfkd03Aud6x54LU5ct/it+xLcDqYM8iUDPcs4BT+tm4Bk9ikPI1CnOvI/o9iqri0rKxT6kG2xYAFwUp0y2fcgp8FeKh+op9ivd1YKV3fGqwh8sr24T9U6oHAaO8h9AnXxYwHxjsrzS8/H28PM+GqO92L72X37EB3rHBQfJ3D/ZnY/+U6odB8l/mpf0UJO1EAsatca1HBb4IcX7ffbgpyPk/CpK/qZc2MsRv36QQv11j77daHHDc96LsGaTMFC+tR5C0gV5a7wjP39HL/2h+14R7Ga8FNpP7hVsNSMUpdPE7vgXXSi0XgRy5ngWginf+X/3rLI7Nuv8Roqri+y4iFYF2uC7RxyLSTlUfDlJsqqp2K+Cp+nibPz8AZ6hqUXT/UdVtOOuCJsBpQCdcF+twb7tRRE5X1d+9Ip29z8YiMiBIlS29zzZA4BBAsKES33jtQYW6gOAEO89a73N2kLQ13mdDv2O+66wa4jp9Y+l5zOpCnL8orvMaXItxeMDx4UA/4Fpc9z4YEcsoIjWAe3FDBM1wrVp/go7f+qOqmSLyDvCoJ9sIL+lSXI/sbfU0osfHwK3AQhH5HNewmK6qOyI4104R+Ro3rDNPREbhepe/qWpqfuX3B1OqhUBVU4CZInIOrntyn4i8qUEmcwrBlao6XJydYTPcmFh/4A3cH6jIUNUVwFvehog0xLWWzwTeAY7wstbwPs/Lp8pKQY5tD3Is0/uMpm1lsD9eZgRpZfyO+a7zFG8LRUTX6SkViNJ1es/IVbgexocByeOB9cCZIlJXVdcHkSePjAT5LbzJvd9xLe2ZuPHJrV7earieSbkIxX4beBi4nn1K9TpgLzAsIO+dwDLc+O4D3pYpIuNwQ2FL8jlXf+B+4CL2jcuni8hI4B5VLRITQZv93w+8h/Jv3MupY5TrzlLVf3EPxG/A1SJyVjTPEYEMq4ELcA98exGp7iX5lFIfVZUwW6A9bUnDd52353OdgZM6xUVv3Bh2ArDaz3JAcZOadXHP5lX7eZ5rcAp1oKoeo6o3qeojqjqA/CeLcqGqa3CTlCd4FgG+CaovVXVTQN4sVX1JVdvjbLb7AV/iJubGi0hYRa6qaao6QFUPARoBlwC/eJ8jCyJ3QTCluv/4uklFci/VzUTf7u0+LcW/UmYPTqkC+IZAZnifxxfxubMgp0UWC4r1Oil4C/Za7/Mb3ERb4DbcS79avCZyIWnhfY4KknZiIep73fu8HtdKBa93FApV3aiqo1X1fOBHoDlOGUeEqq5S1Y9xw1tLgK7ekEbUMaW6H4hIX9wbPAO34qhIUNXfcH+c1rjJlqghIhVF5H8SehnqHbju7UJV3eIdG4OzFLhZRHqFqLeziFTYT/F852u0n/UUClWdhRuHO0dEgrb2ROQwcba++0OBr1NEDsaZBW0DzlPVa4JsV+JaZs3YZ/5XGFZ4n90CZOiAM3kqKJOAf3D2uecDf6vq5IC6y4nIcYEFRaQMzroB3ORWUESklogcFiSpIu55zmRfYyGq2JhqhARMVFTE2R329PYfCjE+0yTEBIePl0KMawXjUeAM4DER+VhVo/VAlMEZjD8mIjNxZkLbcA/ucTgbzhTgBl8BVc3wxpO/B74VkWleuVTgYNwkVzOcTeH+TApMwo3bjvbG0dKA/1Q1cPywKLkI1zJ6T0Ruww3FbMdNaB2Oay11Bjbuxzkm4SaB3vEmVHYB21X1tTBlrsa1bD9St0AkFO/i7DKvw014FoYPPPleEpHuuBV4LXHDD6NxY5cRo6oqIm/iTBPBjbMGUh5n77wEN7H4H2514Sm4icGxqroozGkaAHNF5E+cSeIqnEVAb9ywyCuquqsgckdMcZoalMSN4OZQmTg7zDH42dX5lekWolzg1sSvzHCC2KkG1DvKy3NrwPEmFNKkCtdbOR14Hqcw1uJa3rtwD+NLhDD1wRmJP4Uzrk4FduP+cCNx41ZJfnkHeDJ2C1KPT/7hAccTgSG4yYoMApb+Et6kKs999PtdBkQqg5dWGXgI9+fejVPuy3H+Eq4DKkZyfr/naUqQ43cBi3DDLWF/S+8385naHZ7P71sB9xLYC9T2jk0h9NLroPLjGhFjcS+PFO9eXBPmt/M9z6GenYNwwx5pBLEZxb3s78NZLqzE2URvwg3J3ACUDcif61nATaA9inshrvHu6zrv2i+kCM2sxBPAMAyj2BCRbsBkXEs77GqskoaNqRqGEQvu8z7DDXGUSGxM1TCMYsGbOOoNHInnmlDdJGypwpSqYRjFxZG4MfKdwBc45zqlDhtTNQzDiCI2pmoYhhFFTKkahmFEEVOqRpHieV5fKM5jvorIHcVwzriJrFnSkYBoC0b+mFItJXjOKV4Vkb/EhfvYKyJrxYXwuDo/5xNFJNMFOM/w6bhFBAPZt57+gMJT9D6HJyeFyTfML9+A/Txnt2jUYxQMm/0vBYjIo8BjuJfkdJw39t04zz7dcEsVb8T5SS1Oevs+VXVt2JzRJX4ia+YlE7cS6cfABBGpglsLn0n8/DfbsH9LjQ844uWHMwqJiDyEawGuwjnWyGP3JyK9gbuLWzacWzqKWaGiqkuL83wF5Bucg5Yaus9BjY+LcctKvwTOLnbJgqCqi2MtQ4mjqNa/2lb0G27d9V5vOzSfvHlCUuBaRT/h/IamAX/ivA4Fy7vC2yoCz+LWY+/BuVG7n9xhMAYQwt+Bn9xB19l76VMIWJuOczt4Oc4b2CbckMIqnFOX/sFkDXYPcI6O/8S1vnbivFCdHyRvjoze909xYUDScR7zIwo3EiCT4jznK3BHkDxzvPt6HUF8FACH4HwtzPLuwR6co5G3gYYBeYeH+g3w/C/gt84f5/9hivcsqF89gf4WmuJ8CWwFGgecsyLOf0EWQXw8HCibtVRLNlfiHE98qqp/hcuoLkppDiIyBKdAN+M8sO/GrXIZApwmIqdqXk9YZXBKrD7O0UUmLhLoUzgPQj6n1FO8zyvIGw21sOxXZE1xYZe/x/n/XAz8H65VeC7wmRfJ9aEgRRvjvN0vw3nXr47zyjRGRHpogMu6CPgBp2CvwY0z++Q7EuiAu1fZwQoC5+CciUzGvVz24sL6XIPz8N9JnRNocLHOwL2IprLvN4G8EWnPxSlVX+TXxqGEV9XlInINznh/hIicqKq+aAGv49xTDlDVKaHqKPXEWqvbVvgN5zJOgWsKWK6zV24lflEzccNBX3tpDwWUWeEdHweU9zteG9dy2Q6UCSgzhSDekChcS7XQkTW9Yw/6yZ8UIL/v2roEkVHxCwbopZ3mq6sA99x3jiTgEe97Z7/0N3EtvEY4JRmspdqA4L2IU72ybwQc7xasHr/0K7z0bOD0EHlCedV63Ut70tu/3Nv/kSARXA+kzWb/Szb1vM/VBSznc7g8WP1iF6lrcdyN+5OFiod1m6qm+ZXZiHOBWBVoVUA5CkoG+7zk56CqmyMoexXuT3+X7mtZ+eQf5O0Gu+b/cBFl/c/3Pe6FdHRkYudhGO46roWcQJIXAd+r6spQhVR1jQb0OLzjE3CRfU8rpDxjVHV8AcvchYu2e7+I3IJr+W/Chf4O1dI+IDClemDii6eVZwZaVf/BKemmIlI1IHmHBg+2VhRRQgP5GNd6XCgiT4rI6UHkC4qIVMaFBFmrwSdefPehQ5C0eaqaR5HjrrlQ16uuiz4OON+T7QKcz9Z3wpUTxyUiMlFENolIpl9MqsOIIKJpCGYWtIA6x9j9cb5VX8UNpVymqusKKUOpwZRqycb3ABf0z+RTRqH+AL7j1QKObw+RvyiioQZyp7ftxk02fQdsFpExItIibMnCXy+Ev+b9+f+8g5vYuQjXYl2PG3oJxwu4cd22uPHh53FjsANxLeqyhZQlT6TVCPkH58gcYCEwoZD1lCpMqZZsfvE+C2qX6YsSWjdEer2AfNHG1z0MNVFaLfCA7l9kzVhfbzDG4TzSPwIcAwzzH5YIxIuDdRsuykIrVb1EVe9XFy10AM4SoLAU1qvSA0AX3GRnOwoXr6rUYUq1ZDMMN87YT0TahssYoHTmep/dguRrgYu/tFwjj59VULZ5nwcHOX8VnOlQSLSAkTXVxSJaCjQQkZZBsnT3PudEIHtU8IYUhuLuteIWaISjGe7/OkEDYiuJSEMvPZDCRmnNFy+09OO4EO2Hep8DRaRrtM9V0jClWoJR1RU4m9CyuAB8QVdMiYjPXMbHUO/zERGp5ZcvEXgO91y8VwQiAzlKbjFwnP/LwDv/C7igb/gd36/Imh5Dcbauz/qHvBaRmsD//PIUJ6/gjPxPU9Vl+eRd4X12DZC/Em4oIVirv0ii0YrIQcAnOKV9gbqgl/1xQyIjRKR6uPKlHbNTLeGo6hARScItU/3di2w6i33LVE/ARb6c5Vdmmog8gwtp8ZeIjMRNOPTEtTp+wRn4FyXP4hT3ryLyBc6ovjvOFnY+0N4v7/5G1gT3sugJ9AHme9FZK+BsXGsDz6jqL2HKRx3PauGrCPOuF5FPcZNa80RkAm6s+BTcvZsHHBFQ7G/cEMMFIpKBu28KfKiq/+2H6ENxivo2VZ3nyTdfRO7GhUcZjhuWOTCJtU2XbdHZcMrlVdyY206cYfg6XAv1aoLbN16AU6C7cH/MBcDDQHKQvCsIEeGTEJFSCRO100u/2jvnHtxkyVtAjcBy7GdkTb/jybioqH/hVpDt8q7/wiB5m1BAW9p8fp8VXn1JEeQNZadaAbcIYgn7VpT9X7B75lfmKJw98w7cWHbQFVVhZAlcUXWrd2xMiPyjvfQ7Y/2fiNVmnv8NwzCiiI2pGoZhRBFTqoZhGFHElKphGEYUMaVqGIYRRUypGoZhRJFSbadavsMtZtpQzGyY/kqsRTjgyDYLnmKnWvlECZVmLVXDMIwoYkrVMAwjiphSNQzDiCKmVA3DMKKIKVXDMIwoYkrVMAwjiphSNQzDiCKmVA3DMKKIKVXDMIwoYkrVMAwjisSdUhWR6iIySER+FZF/RaSzd7yGiDwqIq1jLaNhGEYo4mrtv4g0woW3qIOLr9MMLwicqm4RkQuBmrhQvYZhGHFHXClV4GlcHKEjcDGINgakjwHOLGaZDMMwIibeuv+nAK+oi4wZzPXOcoLEijcMw4gX4k2pViRv69SfysUliGEYRmGIN6X6N3BsmPSeuPDChmEYcUm8KdW3gEtE5Cog0TumIlJZRF4EugGvx0o4wzCM/IiriSpVfUNE2gHvAqne4ZFAVdwL4BVV/ShW8hmGYeRHXClVAFW9RUQ+BvoDLXHKdAnwqar+GlPhDMMw8iHulCqAqk4HpsdaDsMwjIISV2OqIjJVRK4TkeqxlsUwDKMwxJVSBeoDbwLrROQbEblQRCrEWijDMIxIiSulqqotgWOA/wPaAx8DG0VkhIj0FpG4HK4wDMPwEVdKFUBVf1fVu4BGQHfgI9xKq7HABhF5K5byGYZhhCPulKoPdUxV1RuAusC1uIm1a2IrmWEYRmjiujstIgcB5wIXAsfjFgTMialQhmEYYYg7pSoiFYG+OEXaAygL/AMMBj5R1X9iJ51hGEZ44kqpisjnQC+gArAGeA0YoarWOjUMo0QQV0oVOAk34z8C+ElVg7n/MwzDiFviTanWVdXMWAthGIZRWOJq9t8UqmEYJZ2YtlRFZDKQDZymqpki8mMExVRVTy5i0QzDMApFrLv/Qu7WcgLBw6gEljEMw4hLYqpUVbVbuP14oXxyGY4/siUd2hxMh9YH06FtIxrVcz5fBr85jifeGlfgOl95+AKuPbcrAP+t3ULrMx4Lmq/rkS3ocWwbOrZtRNOGNahRrRKVypdj265UFi1dx9jJ8xk6ehrpezLylD3+yJZMePf2iGUa9Ma3DHn7uwJfS7wyd84sPv/kY/6YN5dt27ZSqVJlWrZqxVl9+3FazzOCltm1cydzZv/OooUL+HvRQhYtWsCWzZsBePTxIZzZ5+zivIS4Ij0tjTmzf2fxooX8vWghixctYP26dQBcc/1NXHvjLfnWsWXLZj4a/h6//DSVDevXUa5cOZo2b8EZZ/blrLP7IVKwNtMdN1/H9F9/AaDjkUfxxnvvF/zCokysW6q5EJETgEWquilEek2grar+VJxydWrXhDGv3RS1+k7o1JKrz+kSUd47L+tBrxMOzdnfnbqHPRmZ1K5emdrVK3PiUYdwy0XdOevm11myMnd4r4zMTNZv3hm2/orly1K5YjIAsxf+V8AriV9efel5Phj2bs5+5cpV2LVrFzNnTGfmjOlMnDCeJ599kaSk3H+BKZMn8fijDxW3uCWCBX/9yZ233FDo8osWLuD2m65lx/btAFSoUIGU1FTmz53D/Llz+HHiBJ57+TXKlCkbUX3fjPkyR6HGE3GlVIHJwKU4k6pgnOylJYZILzK27khh3uJVzFu0inmLV/P03edQr1bVAtdTPrkMr//vIjKzsvlz0SqObNc4bP7Jvy1m4vRFTJu7lKWrNrE7dQ8A1atWpH/PTgy+rQ9NG9bksxeupdN5Q/C3QpsxfzlNTwmvIEa+dD1nnHgYq9dv44dpiwp8PfHI6C8+y1Gop57ei9vuupc6deqyd+9eJowfxzNDBjHlx4m8+uJz3HnvA3nK16hZk1at29C6TVtat2nHfXfdVtyXELdUqVKFVm3a0qp1W1q1actLzz2V05IPx+5du7j7thvZsX07TZo2Y8Dgp2jT7lAyMvby1aiRvPTcU8yY9gsvPvMU9z38aL71bdm8iZeff5rKlatQo1ZNVixbFo3LiwrxplTza/uXA7KKQxB/fp27hAbd7s91bNBtZxWqroE3n0nzRrV46p3xNKhTLV+l+tqIKUGPb92RwhufTmXP3kz+738X0rZ5PY49vCnT50f+cNWrVZXTjmsLwIdfzyA7u+SbBWdmZvLWG68B0LpNWwY9+SwJCW7YvmzZsvQ+qy970tN56omBfPbpx5x34cU0bLgv6nmv3mcd0F38cBzR8Uh++GlGrmOvv/xCRGU/+mAYWzZvplxyMi++9ib1GzQEoEyZspx3wUWkpOzmjVdf4qvRX3DBJZfRqHGTsPU9M2QQO3fu5KFHH2f8t1+zgvhRqjE3qRKRhiLSRUR8/eGWvv2ArRfOqcqq4pYxWsrm6MOacNOF3fhnxQaeend8VOqc+efynO8N6lQrUNlLzjyGpKREsrOzGf5l6Qi0sHjhArZucS2niy+7Mkeh+tO333lUrlyFrMxMvvtmbK60xMRi7wSVGPbn3nz3zRgATjmtZ45C9ef8Cy+mQoUKZGVlMf7br8PWNfH775jy40Q6HnkUZ53dr9AyFRUxV6rA1cAvwM+4mf9Hve+B2zdAF+D52Ii5f5Qtk8SbAy5GBG4e/Al79kbHJPe4Di1yvi9bnX83zJ/L+3QGYPLMf1i5bmtU5Ik169atzfnetFnzoHkSExNzWkK/TZ9WHGId0Py3YnnOhFaX444PmqdChYq073AkAL/NCP2b7Ni+neeeHkLZsmV58NGBBZ7YKg7iofv/FbAC1/UfCrxN3vhUCqQAc1R1OSWQh67rSZtm9Rg6+ld+mb1kv+pKLleGBnWqcU6PDjx0XU8Afp79L3MWroy4jhM6taR5o1oADP+ydCqW7OzsMGluFGnp0n+LS5wDlqVL9t3jZi1ahszXvEVLpv/6M8uXLQ2Z5/mnn2Db1i3ccPNt+Q4RxIqYK1VVnQ/MBxCRxsAoVf0rtlJFl/atGnLX5T1Yv3knD734VaHqqFOjMismPhk07Zupf3Ldox8WqL4r+rpW6uZtuxnz4/xCyRSP1K/fIOf70iX/0qZtuzx5MjL2snKls3TYvWsXaamplK9gUXuKis2b9lml1KpdJ2Q+X1rK7t2kpqZQoULFXOk/T53M9999S/MWLbn0iquLRtgoEA/d/xxUdWBpU6iJiQm8OeBiypRJ5J5nvmDH7rRC1ZOVrazfvJP1m3eSlr435/ioCXN4+KWv2LYzNeK6qlYqT9+TjwDg03G/k5FZ7HN/RUarNm2pXqMmAB8Me5fMzLzDLJ+N+JiU3btz9lNSdufJY0SP1JSUnO/Jyckh8/mnpabkfp5379rF04MHkpCQwIOPPk5SmTLRFzRKxLylGgwR6QwcCVQjr+JXVR1U7EIVknuuPIUjWh/Mt1P/ZNQPcwtdz+Ztu3OZRzWoXY1rzu3K7ZeexJndD+fOp75g6OhfI6rrgl5HUT7Z2QIOK2Vd/6SkJK65/kaeGTKI5cuWcuetN3LzbXfQvEVLduzYwXffjOX1V18mKakMmZluwYQEmcwy4ouXX3iGTZs2ct4FF3HY4e1jLU5Y4kqpikhV4GvgONwYq7LPzEr9jpUIpdq6WV0evPZ0dqWkc8eTn0W17jUbtzPw9W+Yt3gVnz5/La881J/f/1rBn/+sybfs5V7Xf+Yfy1m4dF1U5YoHzut/EWvXrOGj94cyY9ovzJiW20C8UaPG9DjtdIa+48KdValSJRZiHjBUqLivG5+enk6lSpWC5ktPT/crs284ZuaMaYz9chS169TlxlvvLDpBo0S8vaKfBI4CLgOa45ToaUAr3CTWHCD0oAwgIteJyCwRmZW5eUERixuelx44n3Jly/DMe9+zbWcqFcuXzbUlJSb4ZN53LKlgP8mYH+ezct1WEhMTcsZJw3FE64Z0aOPsMod9Vbpaqf7cfte9vDv8Y3qfdTbNmregTt16tDv0cG685XY++mw0CQnOPKhe/foRr+AxCkfNWrVzvm/auCFkPl9axUqVco2nDnncLeG+5Y67EYHU1JRcW5Y36ZiVnbXvWFbshrTiqqUKnAm8q6ofi0gN71iWqv4LXCsi3+FMqi4PVYGqvo2zIKB8h1tias3epIG7hEG39WHQbX1C5mtUrzqbpzkj6nufHRnS4D8Uazdup1G96jQ7uFa+eS/v68yBd6Wk88X42QU6T0mjfYeOtO/QMWjaooVu6P7w9h2KU6QDkuZ+M/7Llvwb0tTNZyUQmL5uret9PfrgvWHPM3/uHLp3OQqAZ154hRNP6lFomfeHeGup1gLmed99szH+07Lf4MKtGH40ru+U9+6U9LD5ksuVoX/PTgCM+mEOKWl7w+YvrWzZspmZM5zVXq8zQ7/sjOjQqHET6tarB8D0acHX6qelpTJ/rnvJH3NsZH4x4pV4a6luAmoAqOouEUkFWvilV8AFAiwRhPI85ePtgZdw6VnHhvRSlZiYQFZWaFtLgMv6HJvjg+Cn2eFtLvuefAQHVXHvqNKygqqgZGVl8dSgAWRkZNDu0MPp3KVrrEUq9YgIPXv3Ydg7b/LD+O+46tobqd+gQa48Iz/9hNTUVBITEzn9jDNzpf02b2HY+m+8+nLmzP7dvFSFYA5wjN/+JOB2EZmFc6JyKzEKUV2tcnkSE/c17BO8lRwVkstQo5rfQPyejKi1ALsc0ZxHbzqDoaN/5aff/2XNxu05ac0b1eKKvp25/RLnr3vpyk18OPa3sPVdebZrASxYspbf/iiRaygiYvXqVYz9chQn9TiFps1aUK5cObKzs/lz/jzefP1VZs2cQeXKVXhs0JCgK3K2b9sWtN601NRcacnJySSXL19k1xGP7Ny5g2y/F322uu/p6em57k3ZcmVzjYtectmVjP1yJFs2b+auW2/gscFP0aZtOzIy9jL2y9G89forAPQ957y4NeqPlHhTqu8CV4lIsqqmA/cBPwFTcZNWm4G7YyHYjE8fyOlm+3PXFadw1xWn5Ox/OHYG1z32UdTO27VjC7p2dI31tPS97E7bQ8XkclQov6/BPv/v1Zx/59tBfar6aHZwTbp2dGNV739VulupKbt3M+zdtxj2rm92vyqpqak5JlR169Xj2RdfDTm2d0q34N3PZ58azLNPDc7Zv/aGm7kuAh+ipYnL+vfLtRTYx0fvD+Wj94fm7J9xZl8eHTQkZ79S5co8/8ob3H7TtSxftpQrLjqPChUrsnfPnhxb4mM6H8cdQbyGlTTiSqmq6tc4kyrf/t8i0gLojgu78quqBm9GlELmLlrJlQ+/zwmdWtKxbSPq1KhCjaoV2ZORydKVm5i3eBVfTZrH6Ilz83X6cnmfziQkJLBnbwYjvp1ZTFcQG+rXb8A119/E7FkzWb1yJdu3b6NipYo0adKM7iefQr/z+h9wLcx4oE3bdnwyaiwfDnuPX36awsYN6ylfvgLNWjgn1Wf2PSeoA5yShpTmKNCxnv0/ENkw/ZVYi3DAkV2K/8PxSrXyiSE9ucRVS1VEGuWTRYF0YLOW5reBYRgllrhSqjhvVZEoy3QRmQo8rqoz8s1tGIZRTMSbUr0aN8PfGPgY8NkItQIuBJYDw4GWwCXAVBHpoao/F7+ohmEYeYk3pVoXZ4vaUlVzeU0WkQHAr0A5Vb1dRAbjzKseA2KzdMIwDCOAeJtquwF4J1ChAqjqZpzJ1S3e/ibgPZyvAMMwjLgg3pRqbSCco8QkXGvWx2rir7VtGMYBTLwp1T+Am4NZAXhRAW7GixLg0RIofb7rDMMoscRbK+9u4HvgbxH5ln0TVS2BM7zvFwCISDJwMfBtcQtpGIYRirhSqqr6ixeq+nGgJ3COl5SGU7aPeTGt8Jax5o11axiGEUPiSqlCTiDAPiKSgBtjBdioquHdNRmGYcQB8Tam6k9ZnGeqraZQDcMoKcSdUhWRriLyM7ALWAl09Y7XFJFJInJqTAU0DMMIQ1wpVRHpivOhWhdnk5rjtMCzUxXgqthIZxiGkT9xpVSBwcBC4FDgf0HSpwJHF6tEhmEYBSDelGonYLiq7iG4Y5U15Db+NwzDiCviTalmE95LVX0gtZhkMQzDKDDxplR/B84KliAiZXHG/qU3WL1hGCWeeFOqQ4BuIvIBbigA4GAR6Y2LVdXUy2MYhhGXxJXxv6pOEpGLgNdxrVKAobhZ/+3AReaU2jCMeCaulCqAqn4uIt8Ap+LW/CcAS4DvVXV3TIUzDMPIh7hTqgCqmgp8FWs5DMMwCkrMlaqI5DfjH4iqaszlNgzDCEY8KKehRKZUOwIdilgWwzCM/SLmSlVVrwmXLiLtgYE4hboTeKkYxDIMwygUMVeqoRCRdsAA4Gycwf8Q4DlV3R5DsQzDMMISd0pVRFrjlOm5OOfUzwHPBAsGaBiGEW/EjVIVkZa4cNP9gb3AizhluimmghmGYRSAmCtVEWkGPApcBGQCrwFPqeqGmApmGIZRCGKuVIG/cQb+c4GngLVAcxFpHqqAqtr6f8Mw4pJ4UKqJ3mdH4LN88grO/Coxn3yGYRgxIR6U6pWxFsAwDCNaxFypqur7sZbBMAwjWsSb6z/DMIwSjSlVwzCMKGJK1TAMI4qYUjUMw4giplQNwzCiiClVwzCMKCKqBfEPXbJI2VuKLy5OmbDYVhcXNz3b1o21CAccyUlIqDRrqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUSRpFieXESWAwX1zq+q2rwo5DEMw9hfYqpUgakUXKkahmHELTFVqqp6RSzPbxiGEW1sTNUwDCOKxLr7HxQRKQO0AqoRRPGr6k/FLZNhGEYkxJ1SFZFBwO1AxTDZEotJHMMwjAIRV91/EbkbeBj4DLgcEOAB4EZgITAXODVmAhqGYeRDXClV4BrgK1W9FvjOOzZbVd8GjgLKA11jJZxhGEZ+RKxURaSRiFTJJ09lEWm0H/I0BX7wvmd5n2UBVDUd+AC4cj/qNwzDKFIK0lJdjhvrDMdtXr7CspN947y7gEygQUB67f2o3zAMo0gpiFIVbytKFgPtAFQ1C5gNXCYiZUSkPHApsLSIZTAMwyg00Z79rwuk7Ef5McCdIpLsdfcHAWOB7biVV+WBS/ZXSMMwjKIirFIVkcsCDh0R5Bg4E6dGOIX3Z2GFUdXngef99seJyPHAeUA2MNZsVA3DiGdENfTSexHJJrK1+b5hgVTgHFWdUGBBRMoBpwErVPWPgpYPRsreMBdnFAkTFm+ItQgHHD3b1o21CAccyUmhh0Lz6/77ZtoFGAp8heuiB5IFbAGmq+r2gosIwF7gC9xkWFSUqmEYRnETVqmq6vu+7yJyOc6G9IOiEERVVUSWAdWLon7DMIziIOKJKlXtXpSCeLwIPCwiw1R1XTGcr8jpeFjriPN2Oupo3h6a+521ccMGpkyexKyZv7F48SI2bXTd6xo1a3LY4e05u9/5HH3MsVGVuaQy5cuPGT/i7Zz9p76YGjLv8kXzmf7dl6z4+y9Sdm4nuUJF6jVuTqeTenFE1x5By2zduI5nbr4gYnmO7HY65938YOQXUAp57523eeWlnGkS5i/4O2z+3bt38/mnI5gy+UdW/reC3bt3c1D16jRq1JhORx3NxZdeTpUqYc3lY07ESlVEDgLqAUtVdY/f8SuBvrhZ/5dUdeZ+yFMTZ4v6r4iMAZYBaQF5VFWf3I9zFCs1atQMm56ZmcGOHTsAaHvoYbnS1q9fxxmnnoT/uHdy+fKgyto1a1i7Zg3ffzeOPmf345HHHicx8cB1ibBpzUomfjE8orzfffQWU8eMyNlPrliJ9JTdLPlzNkv+nM2f06dw0V0DSEzM/fdISEikUtXwHanMjL2kp+4GoGGLyF+opZEVy5fx1huvRZx/5m8zeODeu9myZTMAZcqUIbl8eTZu2MDGDRuY9ftMup/Uo/QoVWAIbnY/x/heRG4FXmLfRFVfEemkqgsLKc9gv+8XhsijQIlRqj9M+SVs+ofvD+XF554BoO/Z/XKlZWdloaocfUxnep/Vh2OO7Uyt2nXIzs5mxfJlvPbyi0yZPIkxX46iVq3a3HRrfmszSifZ2dmMfONpMjP20uiQdqz8Z0HIvL/9MDZHobY/7iR6XXojVWvUJjNjL/N//ZEx777Egpk/892Hb9L7iltyla1WszaPvPtlWFnGvPcS08d/SZmy5UK2eA8EsrOzeeyRh9izZw/tj+jA/Hlzw+afO2c2t950Penp6Zzc41SuvvY62rY7FBEhLS2NpUv+ZcqPk6hUuVIxXUHhKYhSPQ6YpKr+Lcd7gDXARTgb1Q+Au3Br+AtD00KWK7F8NXoUAEd0PJImTZvlSqtcpSoffzaKNm3b5TqekJBAs+YteP7l17j1xuuY9uvPjPjofa6+7gbKlStXbLLHC9O+G81/f//FEcefQo26DUIq1aysTH74bBgADZoeQv/b/kdCglv/klSmLEd2O52MvXv46p0XmDZ+NJ17nkONOvUjliNj7x7m/TwRgEOPOYHyFSvv55WVXD75+EPmzZtLr95n0qhR47BKNS0tjUceup/09HQuvPhSHnjokVzp5cuX59DDDufQww4varGjQkFWVDXAbwmqiLQFDgZeVdVfVHUk8DVwQmGFUdX/ItkKW3+8MX/eHJYvcwvEzj7n3DzplStXzqNQ/RER+nit29TU1Jy6DiS2bljHhE/epULlqvS+/Jawedcs+4fdO7YCcPyZ5+coVH+OPrk3yRUrkZ2VxdyfCmYZuOC3n0hL2QXAUSf3LlDZ0sTq1at49eWXqFatGvfe/1C++b/5egyrV62iZs1a3Hn3vcUgYdFSEKVaHkj32z8O1xWf6HdsKbnX6hcIEckSkYvCpPcXkaxQ6SUNXyu1UuXK9Dj19ELVUbZc2Zzv2dml5tZEzOi3nmXvnjTOuPwmKlWtFjbv9k3rc77XbtgkaJ6ExERq1TsYgH/nzyqQLL//OA6AmvUa0qzdEQUqW5p4/LH/kZaWyj33PUj16vkb83wz5isATjnt9FLR0ypI938N4D/yfhpuUmm+37GDyDuxVBDy8y2QSCkJFJiamsIP3zvvhqf3PIPy5csXqp7Zv7t5wTJlytCo8YE1ejJz4tcs+XM2LQ47kiNPLNhLKTs7O9+0Dasi9w20ZcNali1wXdxOJ51RIFlKE6O++JzfZkzn2M5dOLNP33zz7927lwUL/gKgbdt2rFu7lnfeeoNffvmJLZu3UKVqFQ499HDO638BJ5zYrWiFjxIFaalOBnqJyC0icg1wFjBeVf2fzubAqv2UKajSFJEKQA/cIoMSz/ffjSM1NRWAvkG6/pGwZvVqRn7xGQCnnt6LSpXifxA/WuzYsolxH75JmbLlOPv6eyIqc1CtejnfQynMzIwMNq9bDUB66m72pkfWRpj14zhUlYTERI488bSIypQ2NmzYwIvPP0NycjL/e+zxiMqsXbOajIwMwA0b9Ovbm1EjP2frli2Ur1CerVu28NPUydx60/UMfPSRXJYw8UpBlOqTwG7gZeBt3FDAAF+i52u1KzCtIAKIyP9EZK+I7MUp1A98+/4bzhXg5cDogtQfr3w5eiQAh7RqTdt2hxa4fHp6OvfdfTvpaWlUO+ggbrvjrmiLGNd8+fbzpKfupsf5V0Q8mVS/Wcsck6ipX40gKyszT55p40ezJ22fT6D0tNR8683OymL2lPEAtO7YmcoH1YhIntLGoIGPsmvXLm646RYaHnxwRGV27tyZ8/2dt94gKakMz73wMtN/n8sv039n/A+TOfU01wsZPeoLPnx/eFGIHlUiVqqquhznlu92nN/UQ1XV35K3BfAWMLyAMvwBfAyMwHX/p3n7/ttHwOvA1cCtBaw/7li65F/++sONmhSmlZqZmcnD99/DooULSEoqwxNPPUut2nWiLWbcMvenCSyeM516TVrQtff5EZdLTEzi5POcP6CNa/7j/ScfZM2yf8jMyGDXti1MHfMJ33/8di77VJH8vV3+M28mO7duAuCokw/Mrv83X4/h56lTaNW6DZdeHrkfef9hmOzsbAYMeoJTTjudMmXKAFCvfn2efu5FWrVyI4/vvfMmmZl5X4bxRIFc/6nqeiCoNa+qzgHmFFQAVR2D509ARBoDg1V1UkHrKUl8Ocq1UsuVK8cZvc8qUNmsrCwefuBeJv84kaSkJIY8/Syduxw4EWZ2bd/K18NfIyEhkX433JvHQD8/Op92Nts2rOenrz/ln/kz+Wd+7rUqNes15LDO3Zk8+kOAiMyiZk76BoAq1WvR6ohjCiRPaWDL5s08+9QQEhMTeWzgIJKSIv9NKlbcF9+zUeMmnHRyXtvehIQELrvyKh5+4D62b9/OwoULOPzw9lGRvSgolD9VEakIHAJUUtWfoyVMNJbCish1wHUAr/zfm1x1zXX7LVc0ycjYy7hvxwJwUo9TqVyA1SFZWVk88uC9/PD9dyQmJjLoyWcKbTVQUhn/8Vuk7trBsaf2oVb9RuwJ6J5nZWbkfPelJSaVIclr+QD0uuxG2h7dld8nfcPqpX+TnppClYNq0KbTcXQ941ymjvkEgGq16uYqF4xd27eyeM50wC1LTTgAV7W99OLzbN++nfP7X0jTps1ITcntUtk3ZgrkpJUpU4YyZctS26+H1bRp6InW5s1b5Hxft3ZN6VGqItIQN6Z6Jvtm4pO8tK64sdabVHVKYYQRkYhsXMP5VPWCBL4N8en6b8qPP7J92zYAzu4Xedff10KdMH4ciYmJDH7yGU47vVdRiRm3bN3ozKJmTBjDjAnBHKbt47HLegJwXK9zOfPK3KNGTVofRpPWhwUrxuqlblSr8SGhbYR9zJn6PdlZWYgInU468H4PcJNNAJ9/9gmff/ZJ2Lydj+4IwMWXXMZ9Dz5M1WrVqF2nDhs3hHcZ6T9BFcmQTCwpyNr/esBvQB2cN/7aQGe/LL95x/oDUwopzxQiM5kqsc2BL0d/AcDBjRpzZKejIyqTlZXFQ/ffk9NCHfzkM5zW88Acuytqdm3fypI/ZwPQMYJZ/FmebWqzdh0KtPrK2EfnLscx5svRLF+2LGSepUuX5Hxv0KBhcYhVaArSUn0MpzRPUdXJIvIYfkpVVTNE5GfcooDCEqz7n4hbvnoDzvt/iXX7s27dWmbOcF3FPmefE9Eb19dC/eH770hKSmLQAdpC9XH9wJfDpv/w+TAmeY5VwnmpCkZ2VhZfvv08WZkZHNyiDYccEf6lt2LRH2xauxI4cCeoAN4b/mHY9Df+71XefN1NxQTzUtX37H6M+XI0K1f+x4+TJuYZV83OzuaDYUMBqF2nTthVhvFAQZRqL1w4k8lh8qwEji+sMKoa8l8gIsNxlgFdgR8Le45YMubLUWRnZ5OUlMSZfc7ON79vDHXC+HEkJSXxxFPPccppB9YYarTZsmEtsyZ9y6HHnkjtho0pU7Yc2dnZrPxnAT98NpSlf80huWIlzrv5gXxfejMnfQtAhUpVOPSYQq/OPuDpeGQnTjn1NH6Y8D0DH32Y7Kwsup10MklJSaxbu5YXnn+Gf/5xyvjW2+4Murw4niiIUq0D/JtPngygYj55CoWqZonICOBeIDLL4jgiOzubsV85D0fHdT2BWrXyj7Q9b+4cvv9unLcnPPPkYJ55cnDI/Pc88NAB3YqNhD2pKUz+8iMmf/kR4Gb396an5disVqtZh0vvHRxyGauP9NQU/pwxBYAjjj+FpDJlw+Y3wvP4E0+xdetWZs/6nbvvvI2yZcuSnFyenTt35OS54aZbOKtv/o2RWFMQpboV50AlHIcA6/PJsz+UB0qkZfVvM6axft1aAPpGOEHlv1gtMzMjx89kKPakp4dNN+Cg2nU5+dzLWbZgHlvWryFl1w7KVahIrfqNOPSY4znm1D6ULZecbz3zf51Exh53v48+gLv+0aJChQq8O+wDvho9im++HsOSf/8lJSWF2nXq0LFjJy68+BKO6NAx1mJGRNjAf7kyiozErfdvqarrvTHVR1U10UtvCSwAPlLVq6IqpFut1R14D1ioqhH1teJx9r+0Y4H/ih8L/Ff8hAv8F3ZwQkQuExGfE8NngWRgqoj0BCp4eSp6+1/jJpKeD1pZBIhItuepKtcGbAO+xDlwubmw9RuGYRQ1+XX/h+PW9/+hqr+JyPXAG8A3fnl8i3czgatUNbTb9fx5nLwmVYpTqkuACap64Pm3MwyjxFDQZapDPbOpm4BjceObO4AZwGsBvgAKjKoO2J/yhmEYsabAy1RV9V/gziKQJRcikoxT2ptUdW9Rn88wDCMaxJ3Bl4h09VrDu3B2r1294zVFZJKInBpTAQ3DMMIQSUu1mog0KkilqrqyMMJ4/gMm4ZTpu8D1fnVuFmeNfRVQsOBBhmEYxUQkSvV2b4uUHCcrhWAwsBA3XlsZP6XqMRXnqNowDCMuiUT57QS2F7EcPjoBD6vqHhEJFhtkDS4UtmEYRlwSiVJ9UVWLa1loNuG9VNUH8o9vYRiGESPibaLqd1xAwTyISFngYgoYA8swDKM4iTelOgToJiIf4IYCAA4Wkd7ATzgXgENiJZxhGEZ+FHZCqUhQ1UkichEuyN/F3uGhuICA24GLVHVGjMQzDMPIl7hSqgCq+rmIfAOcCrTEtaaXAN+r6u6YCmcYhpEPYZWqqsZkeEBVU4GvYnFuwzCM/SHmLVURyW/GPxBV1ZjLbRiGEYx4UE5DiUypdgQ6FLEshmEY+0XMlaqqXhMuXUTaAwNxCnUn8FIxiGUYhlEoYq5UQyEi7XC+XM/GGfwPAZ5T1e0xFMswDCMscadURaQ1TpmeC6QBzwHPqOrWWMplGIYRCXGjVL0YV48B/YG9wIs4ZboppoIZhmEUgJgrVRFpBjwKXIQLyfIa8JSqWgQ5wzBKHDFXqsDfOAP/ucBTwFqguYg0D1VAVW39v2EYcUk8KNVE77Mj8Fk+eQVnfpWYTz7DMIyYEA9K9cpYC2AYhhEtYq5UVfX9WMtgGIYRLeLN9Z9hGEaJxpSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRUS1IdOiSxc707NJ7cXFKmUR7Txc31Y+/P9YiHHCkzXhaQqXZP8AwDCOKmFI1DMOIIqZUDcMwoogpVcMwjChiStUwDCOKmFI1DMOIIqZUDcMwoogpVcMwjChiStUwDCOKmFI1DMOIIqZUDcMwoogpVcMwjChiStUwDCOKmFI1DMOIIqZUDcMwokhSrAUIhoi0ALoDtYGPVXWFiJQF6gLrVXVvTAU0DMMIQVy1VMXxOrAYeAt4HGjmJZcF/gRuiZF4hmEY+RJXShW4H7gBeBrXUs3xrq2qu4HRwNmxEc0wDCN/4k2pXg18oKoPA38FSf8LaFm8IhmGYUROvCnVRsCvYdJ3A9WKRxTDMIyCE29KdTNQL0z64cDqYpLFMAyjwMSbUv0GuF5EagcmiEgn4CpgTLFLZRiGESHxplT/B+zFjZ0+DyhwjYiMBKYBq4DBsRPPMAwjPHGlVFV1I9AJ+BI4Ezf7fwHQA3gf6KKq22InoWEYRnjizvhfVbcA1+OGAWrhFP8mVc2OrWSGYRj5E1ctVRG5UkSq+PZVdZOqbjCFahhGSSGulCrwHrBeREaKSD8RKRdrgQzDMApCvCnVE4BhwPHAF8BGERkuIqeKSLzJahiGkYe4UlSq+ouq3gzUB3rilqX2AcYD60TkNRHpEksZDcMwwhFXStWHqmap6veqeiVQBzgHmAxcCfwUU+EMwzDCEJdKNYByQBVvK4OfkxXDMIx4I+5MqgC8CaozcTaqvYBkYAXwHDAidpIZhmGEJ66Uqoj0winSPkBlYCPwLvCJqk6PpWyGYRiREFdKFbf2fyduguoTYKLZqBqGUZKIN6V6LvCtqu6JtSDRZu6cWXz+ycf8MW8u27ZtpVKlyrRs1Yqz+vbjtJ5nBC0ze9ZMfps+jUUL/mLNmlVs37ad1LRUqlSuQrPmLeh2Ug/69juP5OTkYr6a+GfMV6N57JEH88335jvDOLZzaIOSlSv/44vPPmHar7+wYf06MrOyqFGjBi0PaUXnLsfR/4KLoyl2TClfrgzHd2xGh9YN6NDKbY3qHQTA4Hd/4Il3J4Ys27VDU3occwgdWzekaYPq1KhagUoVyrFtZxqLlm9g7NS/GDpmJul7MkPWkZAg9D/1CC7udSTtD6lP5Qrl2Lw9hV/nLef1z3/lt79WRnQd7Q+pz1V9j6bbkS2oX6sK2aps2LKLeX+v5ZufFvD5D/MLdmMKiKhqkZ4gluxMz46Li3v1pef5YNi7OfuVK1chLS2NzMwMALqd1IMnn32RpKTc77g7b7mBX36emrNfvnwFANLSUnOO1W/QkFdef5vGTZoW5SVETJnE+Jj79CnVhIQEDjqoesh8z77wMh2P7BQ07aMPh/PKi8+zd68LiZZcvjwJIqSmuvtfqXJlfpk+K/rCF5Dqx98flXqO79iMCa9fHzQtP6U66rkr6NW1Tc7+7lTXLqpUYd/6neVrtnDWHUNZsmpznvIVy5fls6cv4+SjnQ/6zMwsdqbsoWqlZBITE8jKyuZ/r3/Hix+HN/554pae3H7hCSR6z+HOlHSSEhOokFwWgPn/rOXYy14OW0ckpM14OuSEeUxbqiJymff1Q1VVv/2wqOoHRShWVBn9xWc5CvXU03tx2133UqdOXfbu3cuE8eN4Zsggpvw4kVdffI47730gV9mjju3MsV26ckSHjjRs1JiKFSsCsH37Nr4f9y2vvvw8a9es5t47b+XTUWNJSIgPhRZP1Klbj+8m/Fjgch++P4znn32KpKQkrrrmevqdex4NGh4MwM4dO/jrrz+YMW1atMWNOVt3pDLvnzXM+3sN8xav5ek7elOvZpV8y03+fQkTf/uHafNXsHT1ZnanuhdR9SoV6H/aEQy+uSdNG9Tgs6cvo9PFLxLYmHv9wX6cfHRLsrKyGfDW97w5chq7U/dSrXJ57rmsG3df2o0ht57BPys38e3Pi4LK8PxdZ3HT+cexK2UPTw6dyIjv5rBh624AalStQNcOTWnbrO5+3qH8iWlLVUSyce79yqvqXm8/P1RVEyOpP9Yt1czMTM44tTtbt2ymdZu2vD/iizyKb9Tnn/LUEwNJTEpi5JhxNPT+uJEweuTnPDnoMQDeHf4x7Tt0jKr8hSHeWqr16jcosFL995+/ufD8fmRmZvDci6/Q45TTikjK6BCtlmpCgpAd8JdZ/OX9NK5XPd+Wan5c1edo/u/BfgCcdN3rTP/jv5y0ds3rMuvjOwF49ZOfue/lb/KU/2DQRZx3SnuWrNrM4ec/l0cpn3LsIYx96Wr2ZmRyyg1vMXNBZEMFhSVcSzXW/4CmQDO/kNNNI9iaBaknLlm8cAFbt7iuzsWXXRm0Jdm333lUrlyFrMxMvvtmbIHqP+zw9jnfN2xYv3/CGjm8985bZGZm0P2kHnGvUKNJoEKNJv5KrkHtqrnSTu/SKuf7ix9PJRgvfDQFgBYH16RL+yZ50h++ugcAb42cXuQKNT9iqlRV9T9V/S9wP78tljIXhHXr1uZ8b9qsedA8iYmJNGrcBIDfphesOzl3zr7xvIYHNyq4gEYe0lJTmfjDBAB6n9knxtKUHo5rv2/Mf9nqLbnSDq7rJsO270pj3eZdQcv/vWIT2dmuI9vjmNyxP5sfXINjDmsMwIjxc6Imc2GJq9l/EVkG3KGqQZtsItIbeEVVS0xr1YfvgQielgXA0qX/5ltPeno6GzesZ9IP3/PuW68D0OHITrRtd2h0BC1lbNu2lQvPP4cVy5eTnZ1FzVq1aN++A2f3O4+jjj4mT/6//vojZwKxTbt2zJ0zi/eHvcf8eXNJSUmhZq1adDrqGC67/EpatDykuC+nRJFcLokGtatyzkmH89BVJwPw85xlzFm8Jmj+hITQiyUTEiSnp9euee5x0S6HO4W9NyOTP5esp1fXNtzc/zg6tGpActkyrN64nYm//cNLI35m5bqi93EfV0oVaAJUCpNeCWhcPKLsP/XrN8j5vnTJv7Rp2y5PnoyMvaxc6Rrfu3ftIi01lfIVKuTKs3nzJnqefELQcxx/YnceGzQkilKXLtLT0li0cAFVqlQlLS2DNatXs2b1asZ9+zV9+p7D/wYMymV18d+KFTnfJ4z/jpdfdON3FSpUICkpibVr1jB2zWjGffM1jzw2kL5n94vBVcUvdapXYsW4/wVN++bnhVw36PM8x32KrkrFZBrVrcbK9dvz5GnnN8FUP2DirGWjmgBs25XG4zeexl2XdANgx+50FKVlo1q0bFSLS3p14uKHP+KHGf8U5tIiJtZjqsEIN7BzJLC9mOTYb1q1aUv1Gu4H/2DYu2Rm5rXR+2zEx6Ts3p2zn5KyO0+exIREqteoSfUaNSlXbp+JSo9TT+e2O++hatVq0Re+hFOrVm2uv/EWPh81hplz/uSnaTOZMWs+wz/8hGOOdXapY74azXPPPJmr3M6dO3O+v/LS8xzSqjUffvIF02bOZdrMuXw44nMOOaQVmZkZDB74KH/++UexXle8k5WtrN+yi/VbdpGWnpFzfNTEP3j4tXFs25mWp8z30//O+X7/lScHrfe+K7vnfK9cMbdddrXK5QGoWbUid13Sjcm/L+GIC56nbo/HqNn9Uc647V1WrttG5Yrl+OiJi2lUt9r+XGK+xFypisitIvKPiPheH8/79gO2jcBdwIQYilsgkpKSuOb6GwFYvmwpd956I4sXLSAjYy+bN2/iw+Hv8X+vvEhSUpmcMhJkMuug6tX5/sef+f7Hn/n5t7l8M2EyV117PT9PncyF5/Zl9Mi8b/8DnS7HdeXGm2/lkFatKVvW2SgmJiZyRIeOvPH2e3Q7yf15P/90BP/9tyKnnP8wTdly5Xj19bc47LDDc44ddnh7Xvm/t0hOTiYzM5P33n6jeC6ohLB5ewpNzxhM0zMGU73bI7Q4cwhPDZtEr65t+P2jO7iqz9F5yixYup5RE93L6ao+R/P07b1pVO8gkhITaHFwTd58+FzO6NqWvRmuUZIdMPPvGzZITExg7cYd9LtnOH+v2AiAqvLjzH+56KGPyM7OpkrFZG678PiivAWxV6q4ZalrvA1cS3RNwLYa+A0YhItfVWI4r/9FXHL5VQDMmPYLl15wLl06tafnySfwyovPUb9+fS678qqc/FWqhLcJFBHq1KnLjbfcweNDniEzM4OnnxjIP38vLtLrKE0kJCRw1z3ODCk7O5ufpkzOSfPZAgP07NWb2rXr5Clft149evbqDcBvM2aQlZVVxBKXXNZs2sHAtyZw5YBPKFsmiVfuO5vDWtTLk++GIV8w+fclANx24fH8/eUD7Pr1Sf784l4uP/Movp66gPG/umd8e0Br17fQAOCt0dNJ25NBILMXrWbK7KUAnHxM0Y6Fx1ypqur7qtpdVbsD/wEP+Pb9tpNU9UxVHaCqO8PVJyLXicgsEZk17L23i+ci8uH2u+7l3eEf0/uss2nWvAV16taj3aGHc+Mtt/PRZ6NJSHBmt/Xq16dMmbIR13tSj1OpW68e2dnZjP1yVFGJXypp1KgxBx3kZp1Xr16Vc9xfiTYLYbEB0Kx5C8CtbtuxfXvRCFmKGDNlASvXbSMxMYErzjoqT/ru1L2ccdu7XP6/EXw9dQH/rtzEirVbmTTzX64e+Bnn3/8BB1Vxcw3/rtqUq+zaTftUgq+FGozFy11aUXf/42qiSlX3e62lqr4NvA2xN/73p32HjiGN8xct/AuAw9t3KHC9tWrXYf26daxaVWIszeKalq1a5Z8Jchufi7n4jYS1m3bSqN5BNGtYI2i6qvL5D/ODrs1PTEzgsJauhTvjj9zP+p9L1kV0ft/PVNTrnWLeUj3Q2bJlMzNnOK+GvQpoF6mqrF2zGoAKFSrmk9vwZ9XKlWzb5madGzRomHO8UaPGOavali1bGrL8sqWuq1qpUiWqVatWdIKWIhrXdz0D/+56pJzRtQ3VKpcnNX0vo3/8M1fa9D9W5NTZqkntkHW0buJ6ISvWbi3w+QtCTJWqiGSLSKaIlPXbz8pnC+3mpoSRlZXFU4MGkJGRQbtDD6dzl645acEsBQL5+qvRbNnsVmwdeVTeCYADlfyWXqsqLz7/DODGV084sVuu9LP6ng3Ad+O+YePGDXnKr1+3jvHffQtA1+NPPOB9LiRGsDT5st6dcnwI/DRnWYHqr1mtIk/e6jy5vTVyOtt35R5TTd+TyUhvouv6czpTvlyZPHUc2aYhJx7pzNvH/RLcd0C0iPXT8Dhu8inTbz+/bVDxi1l4Vq9exeuvvsTiRQvYs8e9TbOzs5k/dw633HANUyZPonLlKjw2aAji142cP3cO1115CeO+HpNnCerK/1bw6kvPM2TwAMCtpup91tnFdk3xztq1a7j4gnMZ+fmnrF61KkfJZmdn88f8edx8wzX8OOkHAPqd158mTXOvJbn0siupV78B6Wlp3HrT9bnMpv788w9uu/l60tPTSU5O5robbi6+CysGqlUuT42qFXK2BC+IcYXkMrmOVyy/b+y/S/sm/PDG9Vx4egca1Mq9BLX5wTUYdNPpvPbAOQAsXbWZD7/N69nr9C6tufn842jaoHrObH75cmU456TDmPLOTTRrWIP5/6zl8XeCG/8MemcC23elUb92VUY+d3lOi1VE6H5UC0YMuYSEhAQ2bN3FK5/8vP83Kgzm+q+I+XvxIi7pf07OfpUqVUlNTc1ZtVO3Xj2effFVWrfJvTBg9u8zueGay3P2y5UrR/kKFUhLS2NPenrO8ZatWvPci69Rv0ED4oF4cKiyZs1qzjhtn71j2bJlqVixIikpKTlu/ICgxv8+li9byvXXXsnGDa6lWsFbkOFz+1ehQgWeeuYFTujWPU/Z4iZaDlVgnwOV/Pjw21lcN+gLIK/LwLT0DHan7aFi+bI5LvfAud07//4Pgq5quqV/V56980wgr9s/cCux+j/wQVA7Vx/HtW/CF89enjOhtX1XGmXLJObIsGHrLs69ZzizFq7O9/ryI5xDFVOqRcyunTsZ8dH7zJ41k9UrV7J9+zYqVKxIkybN6H7yKfQ7rz/J5cvnKZeSksLUyZOYM2smCxcuYMvmzezYsZ2yZcpSo2ZNWrdpS/eTT+HkU04jMTEip13FQjwo1fT0dL4aPZL58+fy9+LFbNu2lV07d1K2bFnq1KlL+yM60OfsfnToeGTYenbt2sWH7w9jyo8TWb16FdnZ2dStV4/OXbpy6eVX5loxF0tirVQrVShL7xPacULHZnRs3ZA6NSpTo2oF9uzNzHEO/dWUPxn9458hnba0aVqHa84+hs7tm9CwdlWqVExm645UZi1axWffz2PUpMgWWdStUZk7Lj6B07u0pmGdagAsW7OFcb8s4rVPf2Hz9pTIbkQ+lBilKiInAB1U9WW/Yxfjuv3VcEH/bo80xEo8KNUDjXhQqgca0VSqRmTEs+u/QAYCx/l2ROQQYCiwF5gJ3ATcGhvRDMMw8ifelGo73MopHxcB6cCxqtoT11K9MhaCGYZhREK8KdUqgL8R2em4iKo7vP2plCAn1YZhHHjEm1Jdj6c0RaQGzivVJL/0SoCFrDYMI26Jq2WqwPfALSKyHeiGU6D+DqvbArGNlWAYhhGGeFOqjwCtgWdxk1N3qepqABFJBs4FSkwkVcMwDjziSqmq6ibgRBGpAqSpqr8PLwG6A6uCFjYMw4gD4kqp+gjm3k9V04C87msMwzDiiHibqEJEqonIUyLyl4ikeNtfIjJERKrFWj7DMIxwxJVSFZEGwFzgPu/Qt96mwAPAHBGpHyPxDMMw8iXeuv9PAnWA3qo6zj9BRHoCI708lwcpaxiGEXPiqqWKM/Z/OVChAqjqd8CrQM9il8owDCNC4k2pVmZfAMBgrPbyGIZhxCXxplT/Bs4VkTxyiUgizk717zylDMMw4oR4U6qvACcAk0Skj4i09ra+wETgeODlcBUYhmHEkriaqFLVoSJSCxgAjPZLEmAP8JCqDouFbIZhGJEQF0pVROoAVwBNgc24FmlzoImXZQXwg6oWbRhEwzCM/STmSlVEWgK/AjVwLVKA+4FLVPXpmAlmGIZRCOJhTPVxnB/VO4FDgbNxFgAviH94UcMwjBJAzFuqOBd/76rqK97+QhHJBr4C2gALYySXYRhGgYmHlmotYHbAsVm4oYBaxS+OYRhG4YkHpZqAm9n3xxecPX5iLxuGYURAPHT/AVqKSBe//areZzsRSQ/MrKrTikcswzCMghEvSvVRbwvkZZyHKh/i7VsL1jCMuCQelKqFnDYMo9QQc6Wqqu/HWgbDMIxoEQ8TVYZhGKUGU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFDGlahiGEUVMqRqGYUQRU6qGYRhRxJSqYRhGFBFVzT+XUeyIyHWq+nas5TiQsHte/JTGe24t1fjlulgLcABi97z4KXX33JSqYRhGFDGlahiGEUVMqcYvpWqcqYRg97z4KXX33CaqDMMwooi1VA3DMKKIKdVShIhMEZEpsZbDyB8RGS4iK2ItRywQkQEiElEXWUSaiIiKyBVFLFbUMKUKiMgV3g+XKSItgqS/G+lDUNSIyBHeQ9ko1rIUFfHwe4hIXxEJFuG31OF3v31bloisF5FPReSQYpThtuI4V1FjSjU3iQQPlR1PHAE8BgRTqqd6W2khlr9H3xieO1YMAi7F2Y6OBs4GpolIvSifZzBQPuDYFUAwpfqfl/fDKMtQZJhSzc1c4CIRaRVrQQqDqu5V1b2xliOKlIjfQ0QSRCQ51nJEgQmq+pGqvqeqNwH3ATVwCi9qqGqmqqZHmFdVNV1Vs6IpQ1FiSjU3TwEZuJZgWETkZBH5UUR2iUiKiEwVkeOD5OsqIr+JSLqIrBSRh0TkSq+b1cQv31kiMlZEVovIHu/zDRGp5pdnADDM2/3Zr7vWzUvPGVMVkTIiskVERoaQ/18R+dlvX0TkRhGZ78m6xev+Nc7vXhQhUf09Qo3liUi3wPsIXA4k+neLvTTfGN8jInK9iCwG9gCne+l3i8jPIrLJ+x0Xi8g9IiL7dSdiw0TvsymAiFzl93xsEpEPRaShfwERqS0ib4nIf971b/Key25+eXL9DuLGlk8Emvvd7xVeWq4xVRHp5+33DhTW+6+piFztd6yOJ89aEdkrIktE5EERKTLdl1RUFZdQ1gJvAbeKyGBVXRgsk4icD3wCTAX+BwjubT5JRHqo6k9evvbABGAbrmu1F7gWSAlS7VVAFvAasAXXzb8aOAzo6uUZDdTDdc8GAf94xxcFVqaqGSIyCrhERCqr6i4/+Y8CWgAv+BV5FbgB+Bh4A6gN3Ar8KiJHqOrmYPeiiInq71EAnsA1OI7DKddgXABU9eTbCqzwjt8FfAuMBDKBU4BngYOAhwsoR6zxjWdvFpEHgCeBX3At2Aa47voJItJBVbd6eb/APbv/BywFqgPHAB2AKSHOc4dX90HAPd6x3SHyfgvsxN3/bwLSLsS94EYBiEhNYAaQjLOHXYv7TYcAjXHPe/RR1QN+w/0BFae86gKpwOd+6e+6W6UAFXFK7+OAOsoDS4Bf/Y6N8X7kpn7HauL+hAo08TteIYhcl3j5jgsma5D8U4ApfvvdvLyXBuR7AdcCrOntd/byXRuQrx3uRfBEKfk9BvjKBeT13adufseGA5lB8jbx8qYCDYOkB/sd38UpiXIB9a+I9bMfcL/P8J7P+sCZuBdFFnA0kA78BCT5lTvDK/eMt1/V2783n/Pl+R28Z3dJmPt9RcC92wWU9zuWBGwEvvQ79hawGagXUOcQIBs4pCjup3X/A1DV9cDrwLkicliQLKfg3r4fiUhN34b7c08EjhWRCiKS6OX9VlWX+9W/GdcaDDxvKuR0w6t4df7qJR9ZyMv5CViDe4Pj1Z8AnA/8oPtan/2BNODrgGvaAPwNnFTI8+830fo9ikC0saq6Ooi8vt8xSUQO8mSZ4skT12PDuJbfJtwzMxbXwrsUaAaUA15S1UxfZlX9FlgI+LriabiXcDfvuouKT4BKOMXv42SglpeGN9xyHjAOyAh4Nr7H9Wa6F4VwplSD8wyuJTIgSJrPxGQc7gH0367H3dMauO5zeeDfIHXkOSYirUVkDK5Fs8Orb5mXXK0wF6Gq2cBnQA8RqeEdPh7XdRsRcE3lgXVBrulQ71piSTR+j2izNNhBEeklIjNwCmarJ4dv5rpaEcgRTe7EvaROwg07NVDVEbjWIsDiIGUW4o25qpskvRtngbJe3FzCQIn+RONEXKv0Ar9jF+D+O197+7VwwwmXkve5mOLlKZLn2sZUg6CqG0Xk/4B7ReSIgGTfi+hqYGWIKjbhftCIEJEquPHAdJwZz784JZIIjGf/Xn4jcON85+K6Qxfi/vBf+eVJALbj3uzBSNuP8+83Ufo9wHUjg5FYCLHy3BMR6YL7U08HbsK1+PYCHYGnif9GzCxV/WV/KlDV10RkLHAWrvV4F/CgiFytqlExi1LVLBH5HLhWRKri/jdnA1+pqu938d3rz3DDL8FYFuL4fmFKNTTP4P4YA9n3pwQ3TgewWVUn5inlISIbcX+8lkGSAw2qu+Pemt1UdapfHcEMrwtk9K6qs0XkH+BCEXkPp1y/VlX/iYAluNbF76q6oyD1FyP79Xt4bAMQkWqqut3veJMgeQuzuOA8nBLtoX4mQyLSrBB1xRMrvM/WuJapP22A5f4HVHUlbsL1NRE5CPeSeZzwtqYFvd+fALfglOkO3Hiuf+9rE25Cq2wEz0VUifc3Z8xQ1S3AK7g3bke/pO9xrbpHRKRcYDkRqeWVzwJ+AM4QkaZ+6TWBiwKKZXufgb/HvUFE81kOVIvkOjw+wXX7r8B1hUcEpH+KG2N6PFjhIh4fi4j9/T08fAq4u196EsFngVNwJlWVCiBmNk455LR8xdmv3lqAOuKRH3ATrrd79wsAEemJm8z82tuvICK5jPpVdRtOKVfL5xwpEeTxr3eaV+8F3rbZk9OXnoWzRDjLs3bJhYhUDva8RANrqYbnOdzbsIPvgKruEpHrcIrqTxH5GGeq0QBnawf7/rSPAacBv3jd1wycSdUK3PCA7+38K+6h+EBEXsV1/XsTfMxnjlfuQW+cdA/wo6puDHMdn3iyvIBTQN/5J6rqLyLyCnCbNxn0HW52tSnQB6d0B4Spv7jY399jAu7evysirXE9icAXnI/Z3uerIjIRyFLVT/ORbyxuXHKiiHwIVMaZZEVk6B6vqOoWcTbST+LM1Eayz6RqJW5oA1wPbLKXvhA3xnkC7j/wRj6nmY1rgDznfd+tql/nU+ZTnAlWBjDcfxLN40GcZcfPIjIU+AM3wdUO12M7jH2t8OgRa3OOeNgIb6Y00EvTgONdcLOlW3F/mhXA58DpAflOAGbilN9/OBu/W7066/jl64QbV92FMxH6CKdUFRgQUOctuC5XJn6mQASYVAWUme3lfTfMfbgMZ9e329sW4ewN25ai36M9ztZyD075DsSN/QWaVJXBjUFvwmuBesebeHkfCSH7xcACT4aVuNb/KUHqH078mVTlud8B+a7GKaY9uEbAR/iZleF6Qa8Af+G63ru9/HeR2xRrQJDfrypu/HObJ8uKgPt9RRB5DvM9C6Fk92R6ETd+uhc3wfUrrheYXBT30/ypxgAReRlnwF9JS9DyO8Mw8sfGVIuYwDEmb4zvUuAnU6iGUfqwMdWiZ4k3zvcvbhzqGtxYW9BJIcMwSjamVIuecbhB8Xq4MdBZwCWq+mvYUoZhlEhsTNUwDCOK2JiqYRhGFDGlahiGEUVMqRqGYUQRU6rGAYfnHX5KwLEB4uf9P94pafIeSJhSNYoEyR2d0xehc7O4kCehloaWaIIpa+PAw0yqjKJmoPdZBuflqA/QXUQ6qepdsRMrD6/h1pKHch9oGBFhStUoUlR1gP++iJyM8yZ0h4i8oqorYiFXIOqiIMQiDpdRyrDuv1GsqOoknAd5AY6C3OODInKR5zF+t3gRNb08FcRFwZwnLlrqbhGZLiIXBjuPiJQVkf+JyFJxUT2Xi8jgUO7ewo1RelEZhorICq+ujeIipt7opV8h+6KDnhgw7DEgoK5jRGSkiKwXF91zlbhon/VDyHWkiIwXFyV2p4hMFJHO4e+yEUuspWrEAl+45sCVJ3fjPDp9DUzGeS5CXJjuH3Eu/+YAQ3ENgtOAESLSTlUfyalcRHAeqvrgwp68BpTFRawNFucqtKAiZ+D8cpbDRWH4BOf3sz3O49gbwDzcMMdjOE9kw/2qmOJX11W4qJ57cG4CV+GcmF8DnCkix6pz8OzL3wUXOqQsLpLuElyk0ine/TDikVi7HbOtdG4Ecc/nHe+Bc6WXDTT2jg3w8qcAHYKUGe6l3xdwPBmn6LKBI/yOX+Tln46fezdcgMClXtqUgLp8MnTzO1YT51V+L3BiELkaBrnmKYH5vLRDvHqW4GI/+aedjIta+qXfMcG16BXoE5D/dva5vOsW7Hy2xW6z7r9RpHjd6gEi8oTnvHg8TmG8pKr/BWR/W1XnBpSvgQvVPUtVn/FPUxey5H6vPn+Lgiu9z4fUL6yJutj0gwog/uVAFeAN9Qtz41dfnmiqYbgRN1l3u6quCahnEq7leqaIVPYOd8FFX/1JVccE1PUaIQIPGrHHuv9GUfOY96m4qAM/A++p6kdB8s4McuwoXHiSPOOTHmW8zzZ+xzriWq/BgthNyVfifRzrfX4XNldk+MZBTwwW3gPnkDwR16Kdzb6QMcGUeZaI/AI0j4JcRpQxpWoUKaoq+efKYX2QY77w0kd5Wyj8Y0lVBbaqakaE5whFNe9zTbhMEeK7jmBxx/zxXUdV73NDiHwFuQ6jGDGlasQTwVym+aK7vqiR27XuAKqLSJkgirVuAeTZ7n02AP4sQLlQMgFUVdWdBchfJ0R6Qa7DKEZsTNWId2biuvLHF6DMHNyz3TVIWrcC1DPD++wZYf5s/CKphqgr0uuY432eGJggIokEvzYjDjClasQ16qLEfgx08uxO8ygtEWkufmHAgWHe5xNeiGhfvurAI0TO+7gAdjeKyAlBztsw4NAW4OAQdb2Gi/r5oogcEqSusiLir3CnAX8DJ4hIn4Dst2DjqXGLdf+NksAtOHvOx4FLvUmaDUB93ATVUcCFuAiz4GxJ+wNnAX+JyBjchNa5wO9EqJBUdbPnp2AkLvTyd7jooFWAw3EK1F+ZTwIuEJGvcS3NDNzs/U+qutizUx0KLBCR8cA/nlyNcC3YTbilvKiqisjVuNVno0TE3071ZJwVxekR3T2jWDGlasQ9qrpTRE7ERaC9COiHs1HdgIv9dSdO+fjyq4icBzyAC798C7AO14J9HBc+OtJzfysinXCmWycDp+LCKC8GngzI7rMfPRnohesJDgR+8ur6SETm4xY5dPfqSsGFyh6JC9Hsf+5fvdbrE+wbgvgNN4RxGqZU4xILp2IYhhFFbEzVMAwjiphSNQzDiCKmVA3DMKKIKVXDMIwoYkrVMAwjiphSNQzDiCKmVA3DMKKIKVXDMIwoYkrVMAwjiphSNQzDiCL/D2KzWqAa6TEnAAAAAElFTkSuQmCC\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"print('\\tClassification Report for BERT:\\n\\n',classification_report(y_test,y_pred_bert, target_names=['Negative', 'Neutral', 'Positive']))","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T19:53:44.471458Z","iopub.execute_input":"2024-09-23T19:53:44.472068Z","iopub.status.idle":"2024-09-23T19:53:44.496969Z","shell.execute_reply.started":"2024-09-23T19:53:44.472031Z","shell.execute_reply":"2024-09-23T19:53:44.496186Z"},"trusted":true},"execution_count":71,"outputs":[{"name":"stdout","text":"\tClassification Report for BERT:\n\n               precision    recall  f1-score   support\n\n    Negative       0.90      0.88      0.89      1632\n     Neutral       0.77      0.81      0.79       615\n    Positive       0.90      0.90      0.90      1545\n\n   micro avg       0.88      0.88      0.88      3792\n   macro avg       0.86      0.86      0.86      3792\nweighted avg       0.88      0.88      0.88      3792\n samples avg       0.88      0.88      0.88      3792\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# MODEL 2: roBERTa","metadata":{}},{"cell_type":"code","source":"# tokenizer to split text into tokens that roBERTa can process\nfrom transformers import RobertaTokenizerFast\ntokenizer_roberta = RobertaTokenizerFast.from_pretrained(\"roberta-base\")","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2024-09-23T19:53:44.498242Z","iopub.execute_input":"2024-09-23T19:53:44.498699Z","iopub.status.idle":"2024-09-23T19:53:46.924625Z","shell.execute_reply.started":"2024-09-23T19:53:44.498660Z","shell.execute_reply":"2024-09-23T19:53:46.923845Z"},"trusted":true},"execution_count":72,"outputs":[{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/878k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"55cd6127569a4b828b6d7215e75ef222"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/446k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"41860ccca0a54a28819eb7526c09b61e"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/1.29M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"e384eb7f6999426d9c56447bdd62918b"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/25.0 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"d1f64651f4dd4294b0d2311501714be0"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading:   0%|          | 0.00/481 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"c24740428bb84624a7e93ad49518d3e7"}},"metadata":{}}]},{"cell_type":"code","source":"# check length of longest tokenized sentence\ntoken_lens = []\n\nfor txt in X_train:\n    tokens = tokenizer_roberta.encode(txt, max_length=512, truncation=True)\n    token_lens.append(len(tokens))\nmax_length=np.max(token_lens)\nmax_length","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:53:46.925854Z","iopub.execute_input":"2024-09-23T19:53:46.926410Z","iopub.status.idle":"2024-09-23T19:53:55.139840Z","shell.execute_reply.started":"2024-09-23T19:53:46.926369Z","shell.execute_reply":"2024-09-23T19:53:55.139044Z"},"trusted":true},"execution_count":73,"outputs":[{"execution_count":73,"output_type":"execute_result","data":{"text/plain":"107"},"metadata":{}}]},{"cell_type":"code","source":"# custom tokenize function to generate tokens that roBERTa can process\nMAX_LEN=128 # max length of tokenizes sequences (max number of tokens given to roBERTa)\ndef tokenize_roberta(data,max_len=MAX_LEN) :\n    input_ids = []\n    attention_masks = []\n    for i in range(len(data)):\n        encoded = tokenizer_roberta.encode_plus(\n            data[i],\n            add_special_tokens=True,\n            max_length=max_len,\n            padding='max_length',\n            return_attention_mask=True\n        )\n        input_ids.append(encoded['input_ids'])\n        attention_masks.append(encoded['attention_mask'])\n    return np.array(input_ids),np.array(attention_masks)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:53:55.146747Z","iopub.execute_input":"2024-09-23T19:53:55.147046Z","iopub.status.idle":"2024-09-23T19:53:55.155022Z","shell.execute_reply.started":"2024-09-23T19:53:55.147010Z","shell.execute_reply":"2024-09-23T19:53:55.154337Z"},"trusted":true},"execution_count":75,"outputs":[]},{"cell_type":"code","source":"# tokenize train, validation, test datasets\ntrain_input_ids, train_attention_masks = tokenize_roberta(X_train, MAX_LEN)\nval_input_ids, val_attention_masks = tokenize_roberta(X_valid, MAX_LEN)\ntest_input_ids, test_attention_masks = tokenize_roberta(X_test, MAX_LEN)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T19:53:55.156083Z","iopub.execute_input":"2024-09-23T19:53:55.156381Z","iopub.status.idle":"2024-09-23T19:54:10.035010Z","shell.execute_reply.started":"2024-09-23T19:53:55.156342Z","shell.execute_reply":"2024-09-23T19:54:10.034255Z"},"trusted":true},"execution_count":76,"outputs":[]},{"cell_type":"code","source":"# function to host pre-trained roBERTa model and attach it to a 3 neurons output layer,\n# necessary to perform the classification of the 3 classes (emotions) of the dataset\ndef create_model(bert_model, max_len=MAX_LEN):\n    \n    opt = tf.keras.optimizers.Adam(learning_rate=1e-5, decay=1e-7)\n    loss = tf.keras.losses.CategoricalCrossentropy()\n    accuracy = tf.keras.metrics.CategoricalAccuracy()\n\n    input_ids = tf.keras.Input(shape=(max_len,),dtype='int32')\n    attention_masks = tf.keras.Input(shape=(max_len,),dtype='int32')\n    output = bert_model([input_ids,attention_masks])\n    output = output[1]\n    output = tf.keras.layers.Dense(3, activation=tf.nn.softmax)(output)\n    model = tf.keras.models.Model(inputs = [input_ids,attention_masks],outputs = output)\n    model.compile(opt, loss=loss, metrics=accuracy)\n    return model","metadata":{"execution":{"iopub.status.busy":"2024-09-23T20:15:18.890973Z","iopub.execute_input":"2024-09-23T20:15:18.891606Z","iopub.status.idle":"2024-09-23T20:15:18.899803Z","shell.execute_reply.started":"2024-09-23T20:15:18.891569Z","shell.execute_reply":"2024-09-23T20:15:18.898908Z"},"trusted":true},"execution_count":82,"outputs":[]},{"cell_type":"code","source":"# define roBERTa model\nfrom transformers import TFRobertaModel\nroberta_model = TFRobertaModel.from_pretrained('roberta-base')","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2024-09-23T20:15:21.963074Z","iopub.execute_input":"2024-09-23T20:15:21.963847Z","iopub.status.idle":"2024-09-23T20:15:23.408875Z","shell.execute_reply.started":"2024-09-23T20:15:21.963812Z","shell.execute_reply":"2024-09-23T20:15:23.407867Z"},"trusted":true},"execution_count":83,"outputs":[{"name":"stderr","text":"Some layers from the model checkpoint at roberta-base were not used when initializing TFRobertaModel: ['lm_head']\n- This IS expected if you are initializing TFRobertaModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n- This IS NOT expected if you are initializing TFRobertaModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\nAll the layers of TFRobertaModel were initialized from the model checkpoint at roberta-base.\nIf your task is similar to the task the model of the checkpoint was trained on, you can already use TFRobertaModel for predictions without further training.\n","output_type":"stream"}]},{"cell_type":"code","source":"model = create_model(roberta_model, MAX_LEN)\nmodel.summary()","metadata":{"execution":{"iopub.status.busy":"2024-09-23T20:15:27.575971Z","iopub.execute_input":"2024-09-23T20:15:27.576718Z","iopub.status.idle":"2024-09-23T20:15:29.150816Z","shell.execute_reply.started":"2024-09-23T20:15:27.576680Z","shell.execute_reply":"2024-09-23T20:15:29.150028Z"},"trusted":true},"execution_count":84,"outputs":[{"name":"stdout","text":"Model: \"model_2\"\n__________________________________________________________________________________________________\nLayer (type)                    Output Shape         Param #     Connected to                     \n==================================================================================================\ninput_5 (InputLayer)            [(None, 128)]        0                                            \n__________________________________________________________________________________________________\ninput_6 (InputLayer)            [(None, 128)]        0                                            \n__________________________________________________________________________________________________\ntf_roberta_model_1 (TFRobertaMo TFBaseModelOutputWit 124645632   input_5[0][0]                    \n                                                                 input_6[0][0]                    \n__________________________________________________________________________________________________\ndense_2 (Dense)                 (None, 3)            2307        tf_roberta_model_1[0][1]         \n==================================================================================================\nTotal params: 124,647,939\nTrainable params: 124,647,939\nNon-trainable params: 0\n__________________________________________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"history_2 = model.fit([train_input_ids,train_attention_masks], y_train, validation_data=([val_input_ids,val_attention_masks], y_valid), epochs=4, batch_size=30)","metadata":{"execution":{"iopub.status.busy":"2024-09-23T20:15:35.319008Z","iopub.execute_input":"2024-09-23T20:15:35.319294Z","iopub.status.idle":"2024-09-23T21:04:49.567227Z","shell.execute_reply.started":"2024-09-23T20:15:35.319260Z","shell.execute_reply":"2024-09-23T21:04:49.566488Z"},"trusted":true},"execution_count":85,"outputs":[{"name":"stdout","text":"Epoch 1/4\n1620/1620 [==============================] - 747s 455ms/step - loss: 0.5737 - categorical_accuracy: 0.7743 - val_loss: 0.4065 - val_categorical_accuracy: 0.8559\nEpoch 2/4\n1620/1620 [==============================] - 736s 454ms/step - loss: 0.3345 - categorical_accuracy: 0.8810 - val_loss: 0.2974 - val_categorical_accuracy: 0.8907\nEpoch 3/4\n1620/1620 [==============================] - 736s 454ms/step - loss: 0.2472 - categorical_accuracy: 0.9114 - val_loss: 0.2742 - val_categorical_accuracy: 0.9063\nEpoch 4/4\n1620/1620 [==============================] - 735s 454ms/step - loss: 0.1885 - categorical_accuracy: 0.9350 - val_loss: 0.2284 - val_categorical_accuracy: 0.9222\n","output_type":"stream"}]},{"cell_type":"code","source":"result_roberta = model.predict([test_input_ids,test_attention_masks])","metadata":{"execution":{"iopub.status.busy":"2024-09-23T21:04:54.411854Z","iopub.execute_input":"2024-09-23T21:04:54.412129Z","iopub.status.idle":"2024-09-23T21:05:15.949079Z","shell.execute_reply.started":"2024-09-23T21:04:54.412102Z","shell.execute_reply":"2024-09-23T21:05:15.948393Z"},"trusted":true},"execution_count":86,"outputs":[]},{"cell_type":"code","source":"# convert model's predicted probability distributions (result_bert)\n# into one-hot encoded labels, where each row represents a single sample\n# and contains a 1 in the column of the predicted class and 0s elsewhere\ny_pred_roberta =  np.zeros_like(result_roberta)\ny_pred_roberta[np.arange(len(y_pred_roberta)), result_roberta.argmax(1)] = 1","metadata":{"execution":{"iopub.status.busy":"2024-09-23T21:05:23.949322Z","iopub.execute_input":"2024-09-23T21:05:23.950074Z","iopub.status.idle":"2024-09-23T21:05:23.954906Z","shell.execute_reply.started":"2024-09-23T21:05:23.950035Z","shell.execute_reply":"2024-09-23T21:05:23.954104Z"},"trusted":true},"execution_count":87,"outputs":[]},{"cell_type":"code","source":"# confusion matrix\nconf_matrix(y_test.argmax(1),y_pred_roberta.argmax(1),'RoBERTa Sentiment Analysis\\nConfusion Matrix')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T21:05:27.419254Z","iopub.execute_input":"2024-09-23T21:05:27.419609Z","iopub.status.idle":"2024-09-23T21:05:27.612685Z","shell.execute_reply.started":"2024-09-23T21:05:27.419569Z","shell.execute_reply":"2024-09-23T21:05:27.611958Z"},"trusted":true},"execution_count":88,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 360x360 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"print('\\tClassification Report for RoBERTa:\\n\\n',classification_report(y_test,y_pred_roberta, target_names=['Negative', 'Neutral', 'Positive']))","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T21:05:30.340111Z","iopub.execute_input":"2024-09-23T21:05:30.340404Z","iopub.status.idle":"2024-09-23T21:05:30.363668Z","shell.execute_reply.started":"2024-09-23T21:05:30.340370Z","shell.execute_reply":"2024-09-23T21:05:30.362902Z"},"trusted":true},"execution_count":89,"outputs":[{"name":"stdout","text":"\tClassification Report for RoBERTa:\n\n               precision    recall  f1-score   support\n\n    Negative       0.89      0.90      0.89      1632\n     Neutral       0.74      0.79      0.76       615\n    Positive       0.91      0.87      0.89      1545\n\n   micro avg       0.87      0.87      0.87      3792\n   macro avg       0.85      0.85      0.85      3792\nweighted avg       0.87      0.87      0.87      3792\n samples avg       0.87      0.87      0.87      3792\n\n","output_type":"stream"}]},{"cell_type":"code","source":"# classification matrix comparison\nfig, ax = plt.subplots(1,2,figsize=(9,5.5))\n\nlabels = ['Negative', 'Neutral', 'Positive']\nplt.suptitle('Sentiment Analysis Comparison\\n Confusion Matrix', fontsize=20)\n\n\nsns.heatmap(confusion_matrix(y_test.argmax(1),y_pred_bert.argmax(1)), annot=True, cmap=\"Blues\", fmt='g', cbar=False, ax=ax[0], annot_kws={\"size\":25})\n\nax[0].set_title('BERT Classifier', fontsize=20)\nax[0].set_yticklabels(labels, fontsize=17);\nax[0].set_xticklabels(labels, fontsize=17);\nax[0].set_ylabel('Test', fontsize=20)\nax[0].set_xlabel('Predicted', fontsize=20)\n\nsns.heatmap(confusion_matrix(y_test.argmax(1),y_pred_roberta.argmax(1)), annot=True, cmap=\"Blues\", fmt='g', cbar=False, ax=ax[1], annot_kws={\"size\":25})\nax[1].set_title('RoBERTa Classifier', fontsize=20)\nax[1].set_yticklabels(labels, fontsize=17);\nax[1].set_xticklabels(labels, fontsize=17);\nax[1].set_ylabel('Test', fontsize=20)\nax[1].set_xlabel('Predicted', fontsize=20)\n\n\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2024-09-23T21:05:40.519030Z","iopub.execute_input":"2024-09-23T21:05:40.520061Z","iopub.status.idle":"2024-09-23T21:05:40.862890Z","shell.execute_reply.started":"2024-09-23T21:05:40.520010Z","shell.execute_reply":"2024-09-23T21:05:40.862198Z"},"trusted":true},"execution_count":90,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 648x396 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]}]}