Here are the two devices that I noticed have changed. I haven’t looked through all my devices, so these two are just the ones I happened to spot. Data science may look at my history for these two devices or for any others they care to review.
The first device is called Oxygen2. The app image below shows how it used to be. Note the flat tails on each run. I selected this particular date range because the magnitude of the tail changed on July 8th. That is curious but is not what caught my attention. The change that occurred later, and what did catch my eye, is when the tail started jiggling around instead of being flat.
The next image shows how Oxygen2 looks now. The tail started getting wiggly on October 25. You may ignore the apparent activity during the night of the 23rd, because that was all one run even though it looks like lots of starts and stops - I think my WiFi router had a bad night and the internet was down a lot. Please let data science know that data for “Oxygen2” is from the same appliance as the integration device named “Oxygen.” Thus they can compare to what the energy usage actually was. Yes, the data reported by the integration does have a wiggly tail, but the wiggles are smaller in magnitude than what native now reports.
The second device that I noticed variation on is my stand-alone Freezer. I do not have that one on an integration, so can’t tell you what it is supposed to look like. Sense had always reported just the startup spike then a flat tail, as shown here. For convenience, I captured a similar date range to what was captured above.
Freezer now reports usage with a wiggly tail. You can barely distinguish it these two images, but when I zoom in on the app I find that the wiggles on one run don’t match the wiggles on the next. My understanding of how Sense works is that it detects ON and OFF events on the local device, then displays usage as time history plot that was generated by the AI engine in the Sense mainframe. This process would not allow detection of variations from one run to the next, so I doubt that this variety really mean anything. I am curious about why the variations began appearing. For both devices, the changes happened on or around October 25.
Justin, thanks for offering to share my case with data science. I’m looking forward to their response!