-
How much days can we use for the prediction od the next day? intuition says 1 week?
-
As the lagest part if the datas are from june, should we discard the rest? or maybe the everything from march and april?
-
to what extend do we do statistical analyis?
-
which statistics for the general analysis?
-
what is the difference between valence aand arousal?
Just take all months of data
avergage per day apps kan je weg laten, maar moet kunnen laten zien dat ze niet significant zijn dmv of wel data analyse of wel referentie
met lege waarden kijken wanneer ze zijn gerapporteerd. begin, eind of tussendoor
- mood: final
- circumplex [arousal, valence]
- countables [call, sms]
- time based [ appCat.all ]
- screen time
- [*] Boxx plot of all variables
- [*] write func that sums call and sms data per hour/day
- [*] write func that sums/means of the appcat data
- decide on what to do with the screen time
finance, weather and game
check the ratio of app usage for only user that have the specific low usage app -> not significant -> discardable
we set them to zero for all users without