Digital Deputyship traffic distribution

As we don't have enough data per day to see usage pattern for the site, then we need to be creative.

What if we import data from last month, and group it over the day of week or an hour of day?



In [282]:

    
import pandas as pd
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline

Let's load data



In [283]:

    
data = pd.read_csv("Analytics All Web Site Data Audience Overview 20150905-20151005.csv", comment="#")
data=data[:-1]
data.describe()



In [284]:

    
hour_start = "20150905"
#hour_start = datetime.datetime.strptime(hour_start,"%Y%m%d")
#seems like pandas is smart enough to process the data string so line above is not needed (for now)
date_index = pd.date_range(hour_start, periods=len(data), freq='H')

Data cleaning



In [285]:

    
#clean up data
del data['Hour Index']
indexed_data = data.set_index(date_index)
indexed_data.plot(title="sessions per hour over time period")









    Out[285]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f4a1be7cd50>



In [286]:

    
# last_line = None
# for index, row in indexed_data.iterrows():
#     last_line = row
# last_line.name.dayofweek
# last_line.name.hour

sum of sessions per week day (0 - Mon)



In [287]:

    
indexed_data.groupby(lambda x: x.dayofweek).mean().plot(kind="bar", title="sessions per week (0-Mon)")









    Out[287]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f4a1bda7d10>

sum of sessions per hour



In [288]:

    
indexed_data.groupby(lambda x: x.hour).mean().plot(kind="bar", title="sessions per hour")









    Out[288]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f4a1bbd3b10>

sessions heatmap

x axis - hour
y axis - day of week



In [289]:

    
heat = indexed_data.groupby(lambda x: (x.dayofweek, x.hour)).mean()
mti = pd.MultiIndex.from_tuples(heat.index)
mti_data = heat.set_index(mti)
unstacked_data = mti_data.unstack(level=-1)
plt.pcolor(unstacked_data.T, cmap=matplotlib.cm.Blues)
plt.gca().invert_yaxis()
plt.title('Sessions heatmap')
plt.xlabel("day of week")
plt.ylabel("hour")









    Out[289]:





<matplotlib.text.Text at 0x7f4a1b974510>



In [ ]:

	Hour Index	Sessions
count	744.000000	744.000000
mean	371.500000	0.489247
std	214.918589	1.032392
min	0.000000	0.000000
25%	185.750000	0.000000
50%	371.500000	0.000000
75%	557.250000	1.000000
max	743.000000	8.000000