Best way to filter Timesheet

时间:2016-08-31 18:28:01

标签: java android python machine-learning data-analysis

I have one application that collects periodically the timestamp from the user and recognize it like outside or inside the place. The problem is that sometimes I have bad precision and some data are collected wrong.

I want to be able to based on current timestamp collected say that this is a expected value of IN/OUT and ignore if don't. I was thinking of a state machine to solve that but I'm not sure how to solve this.

PS: I already have the information if the user is IN or OUT using Geofence

Example:

 IN: 8:28:15 AM (Work IN)
 OUT: 8:29:26 AM (wrong)
 IN: 8:31:54 AM (wrong)
 OUT: 8:33:12 AM (wrong)
 IN: 8:35:34 AM (wrong)
 OUT: 11:54:23 AM (Lunch OUT)
 IN: 11:55:45 AM (wrong)
 OUT: 11:57:11 AM (wrong)
 IN: 12:55:32 AM (Lunch IN)
 OUT: 12:56:11 AM (wrong)
 IN: 12:58:10 AM (wrong)
 OUT: 17:59:03 AM (Work OUT)
 IN: 18:01:23 AM (wrong)
 OUT: 18:03:12 AM (wrong)

Expected:

 IN: 8:28:15 AM (Work IN)
 OUT: 11:54:23 AM (Lunch OUT)
 IN: 12:55:32 AM (Lunch IN)
 OUT: 17:59:03 AM (Work OUT)

0 个答案:

没有答案