This PR addresses several minor bugs identified recently:
Time segments (rows) included in processed phone applications foreground features data were dependent on the order in which app categories and single apps were processed
Phone screen episodes shorter than or longer than specified thresholds were being dropped based on their within-segment rather than total duration. We handle this correctly (i.e., drop based on total duration) for phone applications foreground
Non-Unicode characters in phone Bluetooth bt_name and phone keyboard package_name, before_text, and current_text columns could not be parsed by pd.read_csv(). We add mutation scripts to coerce these columns into UTF-8 character encoding
Minor corrections/updates to RAPIDS installation instructions for Docker and Ubuntu
Tests initially failed because R 4.3.0 was being installed during the tests workflow and this R version was incompatible with some of our package versions. This has now been fixed.
This PR addresses several minor bugs identified recently:
bt_nameand phone keyboardpackage_name,before_text, andcurrent_textcolumns could not be parsed bypd.read_csv(). We add mutation scripts to coerce these columns into UTF-8 character encodingTests initially failed because R 4.3.0 was being installed during the tests workflow and this R version was incompatible with some of our package versions. This has now been fixed.