Feature vector creation raises errors on empty fields in raw data

[sechkova]

Describe the bug

If the raw data .csv files contain empty fields (fields that equal nan values), the automatic feature creation crashes with an error.

Reproduction steps

python ./ml-conversational-analytic-tool/featureVector.py <rawdatafile>

Traceback (most recent call last):
  File "~/workspace/ml-conversational-analytic-tool/ml-conversational-analytic-tool/featureVector.py", line 162, in <module>
    df = featurizer.formFeatures()
  File "~/workspace/ml-conversational-analytic-tool/ml-conversational-analytic-tool/featureVector.py", line 70, in formFeatures
    pull_analyzed = self.commentAnalyzer.analyzeComment(row["Body"])
  File "~/workspace/ml-conversational-analytic-tool/ml-conversational-analytic-tool/commentAnalysis.py", line 31, in analyzeComment
    result['Sentiment'] = self.getSentiment(comment)  # Determine sentiment
  File "~/workspace/ml-conversational-analytic-tool/ml-conversational-analytic-tool/commentAnalysis.py", line 77, in getSentiment
    return self.vader_sentiment.polarity_scores(comment)["compound"]
  File "~/workspace/ml-conversational-analytic-tool/.env/lib/python3.10/site-packages/nltk/sentiment/vader.py", line 361, in polarity_scores
    sentitext = SentiText(
  File "~/workspace/ml-conversational-analytic-tool/.env/lib/python3.10/site-packages/nltk/sentiment/vader.py", line 270, in __init__
    text = str(text.encode("utf-8"))
AttributeError: 'float' object has no attribute 'encode'

Expected behavior

Preprocess the input raw data in order to handle empty/nan fields.