python - Activity log into Pandas Series -

i have input csv (userid, datetime) in order analyze activity of users generate pandas series index being date, , columns being users (multiple series). values sum of activities of each user aggregated on date (for instance: day).

thanks in advance.

fake data

n = 100 np.random.seed(1) userid = np.random.randint(0, 10, n) datetime = np.random.randint(0, 10, n) + pd.timestamp('2016-1-1', freq='d') activity = np.random.randint(0,1000, n) df = pd.dataframe({'userid':userid, 'datetime':datetime, 'activity':activity}) df.head(10)      datetime  userid  activity 0 2016-01-10       5       788 1 2016-01-01       8        44 2 2016-01-03       9       271 3 2016-01-01       5       670 4 2016-01-08       0       475 5 2016-01-02       0       910 6 2016-01-08       1       499 7 2016-01-10       7       787 8 2016-01-09       6       251 9 2016-01-05       9       666 

solution

df.groupby(['datetime', 'userid'])['activity'].sum().unstack(fill_value=0)  userid         0    1     2     3     4     5     6     7     8     9 datetime                                                              2016-01-01     0  166  1091  1878   583   670     0  1524   577   881 2016-01-02   910    0     0   810  2146   706   182   138  1157     0 2016-01-03     0    0     0     0   433     0  1955  1914   566   561 2016-01-04    51  407   598     0     0     0   440   783     0     0 2016-01-05     0  324   662     0     0     0     0   990    79  2849 2016-01-06     0  959     0   230   878     0     0   656   879   300 2016-01-07  1390  100     0   575     0     0     0   806    87  1243 2016-01-08   975  499   503     0   657     0   403   755     0  1271 2016-01-09   342    0   739   617     0  1297   251  1207   324   458 2016-01-10   963  832     0     0   975  1179     0   787   717   145