User analytics



In [1]:

    
from __future__ import print_function
import sys
import numpy as np
sys.path.append('../')



In [2]:

    
import whatstk.parser as wp
import whatstk.vis as wv



In [3]:

    
#chat = wp.WhatsAppChat("../chats/samplechat.txt")
chat = wp.WhatsAppChat("../chats/samplechat.txt")

Number of internventions

Let us have a look at which users tend to have more interventions.



In [4]:

    
# Dataframe containing the cummulative number of user interventions per day (only days of chat activity are considered)
interventions_day = wp.user_interventions(chat, 'days', length=True)
interventions_day_cum = np.cumsum(interventions_day)



In [5]:

    
import matplotlib.pyplot as plt
import matplotlib
matplotlib.rcParams['figure.figsize'] = (20.0, 10.0)
wv.temporal_data(interventions_day_cum, "Cummulative number of interventions")









    



/Users/lucasrodes/Library/Python/3.6/lib/python/site-packages/matplotlib/cbook.py:136: MatplotlibDeprecationWarning: The set_color_cycle attribute was deprecated in version 1.5. Use set_prop_cycle instead.
  warnings.warn(message, mplDeprecation, stacklevel=1)

Now let us take a look at the distribution of the length of the interventions of each user. This can give us some insights about which users tend to send out longer (shorter) messages.



In [6]:

    
hist = wp.histogram_intervention_length(chat)



In [7]:

    
import seaborn as sns
sns.set_style("whitegrid")
ax = sns.violinplot(x="user", y="length", data=hist)
plt.ylim(-10, 100)
plt.show()