Hi @shawn et al -
Are there devices you started using in June that hadn’t been used in other times during the year, like an AC or pool pump? Smaller device signatures can get lost when these big devices are on, but we’re always looking for ways to minimize that. Heaters in the winter (especially radiant floor heating and rooftop/sidewalk ice melters) are especially noisy and can present a lot of challenges, and it’s possible that some warm-weather devices might be having a similar effect. We’ve been doing some research into how seasonality effects device models and detection, so anything you can share about what may have changed would be helpful!
To @gary1’s issue, the data science team made a bunch of progress on ACs last summer with all the data that came in, and are now working hard with the new data that’s coming in which will hopefully mean more people get AC’s detected in the future (the winter is understandably challenging for AC model refinement and testing). It’s possible that there work may have disrupted a previously existing model for you. I might suggest reaching out to support so they can check more carefully. it’s also possible that something in your ACs startup behavior has changed slightly from last summer to this summer, which could indicate a potential problem, but not necessarily. This is the kind of thing that we’re hoping to be able to detect in the future, and maybe even call out like “Hey! We think your AC turned on, but it looks a little different than usual. Did it?” That’s quite a ways down the road, but we do hope to be there. Some of our work towards new integrations may also help this as well. Do you use a connected thermostat by any chance?
We may be able to cover some of the data science teams work around ACs, seasonality, and integrations in a bit more detail in the webinar next month, so please be sure to register!