This PR introduces dtplyr in places that have poor performance. In these places, we wish to calculate summaries from agents' tours per agent which means ~1.3-1.7M groups (agents) with ~1.7 rows (tours) on average per group. As suggested here (https://github.com/tidyverse/dplyr/issues/5017) I used dtplyr instead of dplyr in key functions that I recognized to be slow.
Runtime decreased 98 % from 210 minutes to only 4 minutes.
This PR introduces
dtplyr
in places that have poor performance. In these places, we wish to calculate summaries from agents' tours per agent which means ~1.3-1.7M groups (agents) with ~1.7 rows (tours) on average per group. As suggested here (https://github.com/tidyverse/dplyr/issues/5017) I useddtplyr
instead ofdplyr
in key functions that I recognized to be slow.Runtime decreased 98 % from 210 minutes to only 4 minutes.
Closes #62