I had another device detected and found what it was right away.
It’s a space heater in my garage and so I did some looking close
at the device page because I haven’t been trusting usage data.
Here is what I found:
The pictures are out of order but they were taken back to back
right at 1:10 a.m.
The deice page shows the heater is on and using 1527 watts, it’s
been on for 27 minutes 12 seconds.
So that means it’s been on since around 12:43 so 17 minutes between
12 and 1.
Now look at the next picture which is after I scrolled down and highlighted
the green bar for the 12 to 1 hour and the total on time in the lower right.
How do I have .4 KWH usage in 1 minute 55 seconds? Impossible
The .4 is really close for 17 minutes use but not 2 minutes
Neither one is at 1:10am with proof being your phone’s time on the screenshots. That aside, get some rest and know that these are minor issues. Thanks for your efforts to give support the data.
Did you actually check that the heater was physically on as long as Sense indicated it was?
In the early learning stages (as we have all said many times, you are still in) Sense can sometimes pickup an “On” signature, but can sometimes miss the “off” signature.
In this scenario Sense can show a device still being on (and consuming wattage) for a period of time after it actually turned off.
In this case it’s possible the discrepancy is being caught in the consumption category, just not in the device display. Yet.
Yes, I was watching the timeline as I turned it on and off several times
and it corresponded with that heater every time.
even if it wasn’t that heater but another 1500 watt source. The difference
for two different calculation on that single page cannot be explained the way
i’m thinking about it. As far as I know we don’t have another heat source that
uses 1500 watts except the Keurig coffee maker and it runs for a minute
each time and was correctly detected very early on.
If I had to guess, those stats are calculated at different rates.
Based on your screen shot, I would say that Total usage is calculated on the hour as the hour closes. The .4 kWh seems about right for 17 minutes at 1500w.
I’m thinking that maybe Total Time on calculates by the UOM once a device has turned off. So prior to 1am, the heater had cycled 2x and been on for a total of 1m 55s. It has been on longer (27 minutes) but has not turned off yet in that cycle, so the total on time hasn’t been updated.
I know that from checking my devices, I haven’t seen a total time on change while a device has been on, even if I go to another page and return.
So yes, the discrepancy may be real, but it s only real until the next stats update.
I saw the same about calculations not taking until the full hour
is over.
I’m not following about the 1 in 55 sec. If it hadn’t had to time to update
the calculations then the .4 wouldn’t be there I would think.
I’m not understanding how it could be on for both 1 55 and 27 minutes.
I know for a fact it had been on for 27 minutes as I was right in front of
it the entire time and it had not been turned off during that time. I believe
it showed it had turned on 2x during the 12-1 hour. When I first turned it on
an got a notification, I had immediately turned it off then back on to verify
the detection was identified correctly. So it ran for a few seconds then turned off
for a second. Then back on for the last quarter of the 12-1 hour and continued
to run into the 1-2 hour that had not yet been closed and calculated.
I think the “total” in white text is a total of all KWH during that particular
hour block for either week, month or year. I haven’t done the math or read where
that has been explained yet.
So yes, all of those totals are for the Unit of time you have selected (Day/Month etc…)
Two things going on
with my comment about the 1:55. Without knowing how often / when the Sense calculates that, I can’t know for sure, but I was hypothesizing that maybe it only updates Total time on at the end of a power cycle, or when a device is off and at the end of an hour. So I was suggesting that maybe of the 2x it reported as being on, it had only been on for 1:55 for those 2 times. But that was just speculation.
Devices when they are first reported often come in with some wonky data. Some people get a new device detection, and the ML felt good enough about that detection that even though it was just reported being found NOW, it was able to go back in time and back fill some of the detection from old data.
Other people get a brand new device and it shows zero data. Time times that later gets backfilled. Other-times, the stats start from the moment the device gets added to your monitor.
So its possible that your stats fall into that second scenario where the system added it to your monitor and device list, and included some data for one stat, but not another.
As said by others, take any brand new detections with a grain of salt. That goes both for stats, but also for detections. When my microwave was first detected, it only got every 3rd or 4th usage. But by 2 months later, unless the AirCon is running, it nails it every time.
I see what your saying now and it makes sense. I have detections
that aren’t identified by me yet. The ones that have been confirmed
are not usually hit or miss but where it “combined” something it
found with my oven, that has messed up the oven and the dryer.
The oven and dryer were showing every single time but now, when
the oven is or the dryer is on, they both appear as oven. The dryer
turns on and shows it is off. This is different than merging as I had
nothing to do with it. Sense stated it found a new device and combined
it with the oven. So I’ve really had 17 total detections.
I’ve requested that they change the “combining” to something where
we can have some say in it. I hope I don’t have to delete the oven and dryer
over it.
When I read how long it typically takes for detections, deleting worries me.
I’m confused what you are calling combining. It is just not something I’ve seen before. Did you get a message on your timeline saying “a new device has been found and combined with Fridge” or now when you use the Dryer, the Oven turns on. I assume you still have Dryer as a Device in your device list?
If that is the case, I wouldn’t say that is an intentional combination. Thats the algorithm making adjustments to the model and it made a bad adjustment. My theory, not fact - I think some of the models are very precise. Other models allow for wiggle room, and I would think this is particularly true in restive loads.
For now, your best bet is to report it not on. If the Dryer is on and its showing as Oven, click the Oven device and report as not on. I would say don’t go crazy tagging it every time it happens, but every now and then. I don’t know if those reports go to an automated system that is able to start checking things, or if its just opens another support ticket that gets forward to the Data team therefor reporting it 20 times the same day doesn’t really make a difference because its waiting for a human to look at it anyway.
When I had received a notification about a new device detection
there was the sense stated that it had “combined” it with my oven.
I had not heard the term “combine” before this but is exactly what
it said. I wish that I would have taken a screenshot because when I
read it, it then disappeared for good. At the time I was thinking of
“merge”. This was not merging because my oven page doesn’t
show but the single oven device.
I did not report a problem because I have an open ticket already that
is being worked on and we communicate as needed through email. The
tech that’s helping me was not in today but said he would get back to me.
I think what sense did was assume that whatever it found was a component
of the oven so combined them at a level we are not able to see. The only
difference now is my dryer device is still in the list but doesn’t work.Dryer
on produces oven notification.
I can confirm the report a problem tickets are first an automated response
by email but is absolutely human read within a few days.
