Solved! Go to Solution.
Yeah, some of the more complicated data can be difficult to shoe-horn in but I usually take the approach of looking at the data, trying to create, if needed, a couple of initial views at feature service level, then use those view layers in dash or expb.
For example, with IoT sensors that churn out data every 10 mins, datasets can grow quickly - so where I want to analyse a month, I simply create that view - means less load when creating a graph or widget etc.
Have to admit, not being able to simply (without post processing) invert a chart is frustrating esp when looking at river levels.
So, yeah, we're trying to do similar things, I do like your site - I also work closely with out Flood Risk Management Team - and I always think what we present must be in bitesize, consumable, chunks.