spanishkda.blogg.se - Download nest learning thermostat

I'm happy to find someone who has had the same experience I have! The cycle data is definitely incomplete and flaky. If you were even more of a data-analysis geek than me, you might even be able to process the temperature sensor data to pull out heating times. So, all-in-all, it looks as if about 1/3rd of my 'cycles' dataset is rubbish, but maybe 2/3rds is OK. I suspect some of the heating on/off events weren't collected or processed properly. starting around 6-7am and running for a few hours at most), so the cycles data for those days is also clearly duff.

Again, for the days I checked, the temperature logs show a normal pattern of heating (e.g.

For 6% of the days the 'cycles' data reports basically continual heating all day (more than 16h), often starting at midnight.

For the days I (manually) checked the temperature logs, it was clear that the heating was actuallly running as usual.

For around 25% of the winter days, the 'cycles' data claims there was no heating at all.

From the temperature data it is easy to see when the heating is running, so that's an independent way of checking the cycles data. The data dump includes temperature (as well as humidity, motion sensor, ambient light and some other stuff I don't understand). The two main issues appear I've spotted so far seem to be associated with missing heating data.

my plan is to keep an eye on the 10 day-website history for a while and then do another data request and see if I can do some more data analysis.Ī bit bloody useless though from Nest/google.Īfter a bit more geeking out, I can confirm that the 'cycles' data is probably mostly correct, but is incomplete and somewhat unreliable (in my case at least).

That data from the 'cycles' data structure does mostly agree with the Nest website view, except for a couple of days where the 'cycle' data was blank (and another where I don't believe the figure in the cycle data).Īs an aside there is an 'energy_latest' json file in my download, which appears to be the 10 days of data as displayed on the Nest Website at time of download - that does have apparently correct data in the " totalHeatingSeconds" field - but disagrees with the 'cycles' data on a couple of the days. There is data in the 'cycles' tree - this does seem to be useful - it (claims to) include start and end-times for heating both in time format and via a 'plainText' field which contains entries such as " heating cycle from 6:01 AM to 8:09 AM". Unfortunately this appears largely useless - it's not consistently populated and what data is there seems unrealistic.

The top-level of the data hierarchy is the day.Īt this per-day level there's a parameter called " totalHeatingSeconds". The json files claiming to contain the relevant data are organised as monthly summaries. I've just spend several hours parsing the json files in Python and found inconsistencies between my analysis and the 'last 10 days' data available from Nest.