However, the graphs for the user metrics were grouped by the "week" from the datetime, which was problematic for programs that have already run for over a year. In fact, the most interesting graphs are likely to be year-over-year comparisons, and these would be elided completely by this grouping.
I rewrote to plot by datetime instead, grouping by frequency = Month. But the plots still not look visually appealing, either as bar or as line graphs.
Finally, rewrote as subgraphs which seem to work better for our needs
@zackAemmer made some changes in https://github.com/e-mission/em-public-dashboard/pull/45
However, the graphs for the user metrics were grouped by the "week" from the datetime, which was problematic for programs that have already run for over a year. In fact, the most interesting graphs are likely to be year-over-year comparisons, and these would be elided completely by this grouping.
I rewrote to plot by datetime instead, grouping by frequency = Month. But the plots still not look visually appealing, either as bar or as line graphs.
Finally, rewrote as subgraphs which seem to work better for our needs