We are developing an app that requires background location tracking at intervals... we are, somewhat obviously in designing the app, finding the correct balance between benefits to the user of more accurate tracking, given our user case, against the costs of battery usage. We have I believe, as have a few other Apps we have seen which are already live, found a satisfactory compromise...
IN STEPS APPLE'S MACHINE LEARNING!!!
The app will work for a couple of weeks, and then machine learning will step in for a few days and throttle the frequency of background checks... now get this...
There seems to be A STRONG CORRELATION between these periods and battery draining attributed to the app via the phone analytics... increasing A LOT.... ie 10x
Apple's machine learning is I presume designed to protect the users from too much battery drainage... from background tasks...
So what we have when this occurs is this machine learning function apparently draining the phone of much more battery than the original function it is seeking to improve... and in the process the performance of the original function is greatly reduced.
Without Machine Learning interfering... battery draining appears to be sustainable and insignificant.
Has anyone else encountered this. Apple do you have any comment?
