catherinelee274 / fraud_detection

Predicting whether a user will download an app after clicking a mobile app advertisement.
0 stars 0 forks source link

[WIP] [14] parse click_time into seconds #20

Closed amitSA closed 5 years ago

amitSA commented 5 years ago

parsed click_time to create 2 new feature columns you can work with

click_time_epoch = # of seconds since the epoch (1 January 1970) click_time_secs = # of seconds since midnight of this click_times day click_time_mins = # of minutes since midnight of this click_times day

I would only see the later two engineered features to be useful since the model can correlate certain times in a day with events

fixes #14