I’m not quite a week in using Sense and after the initial quick discovery of two refrigerators and the refrigerator incandescent light, it’s been quiet. This morning my Laser Printer was discovered. Mild serotonin boost. LOL.
@Offthewall I still do as well!
I recently moved out of my old home and I’m currently in the process of figuring out approvals from building management to get it into my new apartment. I have central AC now and am very looking forward to having some visibility into what’s driving my higher-than-anticipated electricity bills .
Ditto as well
Justin / glencsutton… Your landlord has tapped into you panel… End no more… 2 cents… Gerry
I’m an apartment developer, owner, and manger, 200+ units at a time. What if each unit came with a Sense and once the resident moves in, they get private visibility…?
We can start a new thread but I’m a huge fan of Sense and would like everyone, even if they rent, to see their energy usage. I’d like to get the community’s thoughts on this.
All depends how the apartments are setup. If they pay their own electrical then that would be very cool stuff for them. Anywhere where the electrical is included in rent - they won’t care I’m sure. Great idea really.
Thanks, each unit has separately metered electric.
And, while this seems like a great “fully equipped” effort, success will depend on whether your apartments have devices (modern refrigerators, heat pumps, mini-split A/C, etc) that use most of the power and that Sense is incapable of detecting…that just leads to endless frustration
@afreedman It would be nice to be able to provision these and then un-provision once the tenant moved out.
I have a similar case where I have 5 sense devices and I would like to associate a master email that can view and pull data from all of the devices during my study.
Home Assistant works well for acquiring data from multiple Sense accounts.
@kevin1 Thanks for the info! I’ll have to check that out.
Got my high for Sunday. LOL. Garage Door Opener found this afternoon!
I wonder if the detection is “purposely” slow, to keep us interested. If it found them all on the first day, which I don’t know why it couldn’t, we probably wouldn’t be so excited when a new device is detected. BTW, my garage door was the last detection I’ve had in a couple days. I’ve had it over a week and it isn’t detecting my AC or the little power blip that runs every 5 minutes like clock work. (FWIW, I know it is my fridge through circuit breaker deduction) Does anyone know if blink detects lights? I have the same daily routine when it comes to lights and electronics, no detection yet.
I’ve had Sense for more than two years, and it’s still finding things, not quickly for sure and not that reliably, but it keeps chugging along.
The slowness is definitely not a ploy to keep customers interested, it’s the fact that this is a horribly complex (and for many devices impossible) process.
Andy, I saw posts from you from back in 2018. Glad you stuck around. I’m going to share my experiences after a month, 3 months, etc. It seems like they could use some simple logic rather than complex. For example, I get that LED bulbs are hard to detect. But I get up within 15 minutes +/- same time every day, and first light that comes on is a strip of 5 LED in the bathroom. I don’t know why they won’t let us users assign things like that, but machine learning could figure that one out and at least ask me by saying, “Hey, we noticed that what looks like LED lights to us, come on at 5am each day, are these your bathroom lights?” Also, I’ve given them permission to monitor my network, why not look for messages to my wireless printer and detect the power signature based on the pre-broadcast? Or every 8am the roomba vacuums two rooms and 16 minutes later, empties the bag, that should be a no brainer. Or the garbage disposal, within 5-15 minutes of the last use of stove? Okay, I’m ranting, I’ll save it for my 1 month post. Thanks for the reply.
There are very newly-evolving forms of time-series machine learning that are able to look both at the details of a data waveform and the periodicity (daily, hourly, etc) to forecast (and classify). Lot’s of smart home enthusiasts like to think in terms of logic, but logic isn’t so suitable for classification using data that has some unknown degree of variability.
This paper highlights new improvements in time series analysis using a Transformer model architecture, the type that might be able to leverage regularity of behaviors during a daily cycle.
Great research paper, @kevin1. I shared with George and Mohammad. We have a monthly literature club where we catch up on all the latest research and this fits in perfectly.
Keurig coffee maker found this morning. Woo Hoo! LOL
Sense was busy this morning. It found the Microwave in the kitchen and the refrigerator in my RV.