Closer to the End Goal
One of my end goals was to get a better view of my big ticket device over the year. With a little more knowledge of my Exported Data, I’m ready to do that by charting my weekly usage using a program I wrote a couple years ago and have since improved.
Unfortunately, it looks like my Exported data is somewhat less amenable to being plotted than it was a couple years ago. Two things stand out and distort/obscure the results.
-
Kitchen Overheads - There’s a huge Oct power spike for that device that overshadows all the other data. This spike is explainable and shows up in my Sense app data as well - It’s from some buggy firmware in Oct that failed to filter false half million watt surges “measured” by my Wiser smart switch.
-
Other is negative every week - At least according to the Export data. This is where I have to do some digging. Three important things to realize:
- Sense doesn’t output values for Other via Export. A user has to calculate them as Total Usage - all detected usage (Always On + native detections + smart plugs).
- Some of the detections in the Export data might be extraneous - I found a second redundant AC 3 device that nearly matched my main one in the Export file. Extra false loads like this can create negative Other.
- I’m not sure how the Sense app treats negative Other. Places where Other might have gone negative are visible in the Hourly, Weekly and Monthly trend bars - the bar drops to zero. That might indicate that the Sense app zeros out Other any time it goes negative with the app or is used for their data aggregation
I’m going to do three fixes to the data based on these discoveries and see how the results change:
- Filtering out the extra ‘AC 3’ entry
- Clamping the ‘Kitchen Overheads’ to 105Wh when they are on.
- And the same trick Sense seems to do - clamping Other on the bottom to 0.
And voila ! A chart that looks reasonable.
Next step - let’s see how well the data matches the Sense app.