Feed back to data scientists?


@BenAtSense When / how should we provide feed back to the data scientists around device detection, specifically miss identification?

For example I’m seeing my Sense at times miss identify my toaster over & dryer as my stove. The dryer and stove are discovered devices, the toaster over is not identified yet.


I see the same thing on my refrigerator. Sense every now and then confuses a couple things. I would like to see a check and x or thumbs up and thumbs down on the device while it is running. So if my discovered device, Stove, says it’s on, but i know it is not, then I can click the X or Thumbs down. Essentially allowing me to create data points for the data scientist to analyze. Once you have enought thumbs down or X’s then maybe it could help to identify or at least separate the device out of the Stoves usage.


I was thinking of posting a similar question the other day. @BenAtSense, would it be possible for you or Tech support to write up a little guide of basic experiences / expectations. I know that as far as general detection goes, it seems to vary greatly about how long it takes to discover certain devices, and I don’t expect you to be able to produce a timeline, but some general ground rules might be nice.

Knowing that Tech Support seems to be overwhelmed, I have held back from filing any tickets other than major errors (CT issues, Wifi issues, system read out issues etc…)
Is it helpful to the team to get feedback about when devices are behaving incorrectly? I feel like the generic response is always “the system is still learning so give it time” but it would be a little useful to have some guidelines of when it maybe has been given enough time and would be good to file a support ticket. Because as you mention, the system is still learning, but how does it learn without feedback? I think @jglass’s suggestion is a good one. We aren’t “training” by turning things on and off as it has been made abundantly clear that Sense does not feel that is useful, but if we could flag things as incorrect, I feel like that can only be useful.

Just a thought.



I am in the same boat - not filing tickets because the likelihood of them being addressed is low (which is what you’d expect people to do when support is slow to respond and generally are not able to fix problems…i.e. The daytime dip ;)) I’ve got a handful of device detection errors to report, but what’s the point? How can we as users help improve the product without providing feedback?


Completely agree that ways to give more device detection feedback is important. We are working on some features that will allow you to provide more feedback to the Data Science team around device detection. I can’t give an exact date, but they are coming soon and I will be sure post about them when they are released!

I’m sorry you haven’t felt that support has been responsive! The support team has been growing recently, and is working to improve, in order to better meet user’s needs.


It’s been a typical progression for most of us, I think. Some go through it quicker than others. Starts out with optimism and willingness to help by providing feedback. Next comes frustration with problems going unresolved, followed by despair about whether issues will ever be addressed. Last is acceptance, where I think a number of us on the forum are - we’ve realized we can’t do much to help and we just have to sit and watch, hoping some update will come that will solve whatever problem we’re experiencing.


The new update released today gives users the ability to provide our Data Science team a little more feedback via the ‘Device is not on’ and ‘Delete this device’ features. We will continue to work on ways for users to be involved in improving the device process!

As always, thanks for feedback. It’s super important to us as we look at where we can improve.

closed #8