Right now i'm using dates (without times) in my simulated data, but that can lead to false-positive conversions

shane-kercheval / shiny-ab-testing

A shiny app for analyzing AB tests

3 stars 1 forks source link

Right now i'm using dates (without times) in my simulated data, but that can lead to false-positive conversions #1

Closed shane-kercheval closed 5 years ago

shane-kercheval commented 5 years ago

For example, if someone converted at 12pm, and joined the experiment on 1pm, and we are using dates, not times, then we would count that as a conversion, when it's not.

I don't believe the code will break if date/times are used, but I haven't tested; I should add random times to the simulated datasets.

[ ] when calculated if people converted within the time window, add unit test to ensure the 12pm/1pm scenario is correctly identified as a negative number (e.g. don't round)

shane-kercheval commented 5 years ago

this is been fixed I now generate/simulate times in the simulation datasets