human activites.py

# To store data
import pandas as pd

# To do linear algebra
import numpy as np

# To create plots
from matplotlib.colors import rgb2hex
from matplotlib.cm import get_cmap
import matplotlib.pyplot as plt

# To create nicer plots
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE

from sklearn.preprocessing import StandardScaler


# Load datasets
train_df = pd.read_csv('train.csv')
test_df = pd.read_csv('test.csv')

# Combine boths dataframes
train_df['Data'] = 'Train'
test_df['Data'] = 'Test'
both_df = pd.concat([train_df, test_df], axis=0).reset_index(drop=True)
both_df['subject'] = '#' + both_df['subject'].astype(str)

# Create label
label = both_df.pop('Activity')

print('Shape Train:\t{}'.format(train_df.shape))
print('Shape Test:\t{}\n'.format(test_df.shape))

train_df.head()

# Create datasets
tsne_data = both_df.copy()
data_data = tsne_data.pop('Data')
subject_data = tsne_data.pop('subject')

# Scale data
scl = StandardScaler()
tsne_data = scl.fit_transform(tsne_data)

# Reduce dimensions (speed up)
pca = PCA(n_components=0.9, random_state=3)
tsne_data = pca.fit_transform(tsne_data)

# Transform data
tsne = TSNE(random_state=3)
tsne_transformed = tsne.fit_transform(tsne_data)


# Create subplots
fig, axarr = plt.subplots(2, 1, figsize=(15,10))

### Plot Activities
# Get colors
n = label.unique().shape[0]
colormap = get_cmap('viridis')
colors = [rgb2hex(colormap(col)) for col in np.arange(0, 1.01, 1/(n-1))]

# Plot each activity
for i, group in enumerate(label_counts.index):
    # Mask to separate sets
    mask = (label==group).values
    axarr[0].scatter(x=tsne_transformed[mask][:,0], y=tsne_transformed[mask][:,1], c=colors[i], alpha=0.5, label=group)
axarr[0].set_title('TSNE: Activity Visualisation')
axarr[0].legend()


### Plot Subjects
# Get colors
n = subject_data.unique().shape[0]
colormap = get_cmap('gist_ncar')
colors = [rgb2hex(colormap(col)) for col in np.arange(0, 1.01, 1/(n-1))]

# Plot each participant
for i, group in enumerate(subject_data.unique()):
    # Mask to separate sets
    mask = (subject_data==group).values
    axarr[1].scatter(x=tsne_transformed[mask][:,0], y=tsne_transformed[mask][:,1], c=colors[i], alpha=0.5, label=group)

axarr[1].set_title('TSNE: Participant Visualisation')
plt.show()