@kdebrab created a notebook for water leak detection. I checked it today, it's still very useful.
I wanted to create an issue to discuss a few points and remind us of finalising it.
Some remarks/questions:
I changed the vmin in plotting.carpet() from 1 to 0.1. This is the new line of code : vmin = max(0.1, kwargs.pop('vmin', ts[ts>0].min())) It's much more visual to detect small leaks now.
Tthe graphs (based on start and end dates) did not correspond to the explanation. I played around a bit with the dates (later dates give more sensors). We still have bring the dates and the text in line.
For the last graphs, all data is divided by 60. I'm not sure if this is linked with the sampling and units? The units are liter/min, and we should keep that. But when I removed the division by 60, the graphs didn't look good. Let's check on a next developer meeting.
I know that alerting is NOT part of the current sprint, but I thought there might be some nice graphs in this notebook. Let's see next meeting what to do with them.
@kdebrab created a notebook for water leak detection. I checked it today, it's still very useful. I wanted to create an issue to discuss a few points and remind us of finalising it.
Some remarks/questions:
vmin = max(0.1, kwargs.pop('vmin', ts[ts>0].min()))It's much more visual to detect small leaks now.I know that alerting is NOT part of the current sprint, but I thought there might be some nice graphs in this notebook. Let's see next meeting what to do with them.