Daily and weekly cycle averages?


#1

A feature that would be quite useful would be to have graphs in the App / Web for the daily average power usage, and also the weekly cycle. Human behavior varies strongly with time-of-day and day-of-week, and meteorology varies with time-of-day too. So averaging all days on top of each other (and all Mondays, all Tuesdays etc. on top of each other) would produce a less noisy dataset that we could mine for patterns.

There could be different ways to displays this info. The format of the power meter would be good to start. Eventually a box-whisker plot like the one attached (for some research in my field) would be even more useful.

Eventually one would want to select a time window over which to see the averages.


#2

I agree that that would be a cool feature ! If you want it now and know a little R, you can do it from export today. Can you tell when I do most of my car charging ??

Here’s hourly solar and usage by time of day for me…


#3

Here’s a little code to go with that - uses Sense hourly export and can combine multiple years…

SimplePlot.R (3.2 KB)


#4

My time in quant methods courses have made me somewhat stats averse, but your posts really make me want to learn some R.


#5

Stats averse because of the math and essential weirdness of some of the formulations (the null hypothesis), or because of treacherousness of conclusions and causality?

I’m currently finishing a class on probability for machine learning. Ugly to see some of the math and proofs again after so many years, but fun to see what it can do. Problem sets can be done in either Python or R, but R has more statistics built into the base language.


#6

That’s a long tangent that I won’t bore you with, but the short version: My stats experience was all in the social sciences during grad school. I didn’t feel quantitative methods to be particularly applicable to my area of study, so I mostly tuned a lot of it out. I suppose you could classify that as “because of the treacherousness of conclusions and causality,” but this doesn’t need to become a humanities/sciences debate :shushing_face:. Though, I do remember finding SPSS pretty fun. I’ll have to dig into R one of these weekends…as if I need another hobby.


#7

@RyanAtSense Any “sense” of whether this may be in the implementation pathway within let’s say 1 year? I don’t use R but another similar language (Igor Pro), and the R code from @kevin1 doesn’t seem easily translatable unfortunately. Time is short, so if Sense may implement this within 6 mo or a year, I may just wait for it. But if this is unlikely to be implemented, I may invest the time to do it myself.


Any Igor Pro users in the Sense community?
#8

I agree!

If it seems that developing that might be far off, perhaps a nice first step would be to include these graphs in the monthly report?


#9

We have a pretty full roadmap for a while, but I’ll pass this onto the team. I would suggest trying to find a good workaround yourself for the time being.


#10

Here’s another use case for where this would be useful. I set a goal of 25 kWh per day and we killed that goal several days in a row. Then, yesterday I looked when I woke up and we were already trending to exceed the goal. I wondered, How could that possibly be if everyone was asleep?

It was Sunday, so we were in church all day (only 1 trailer renter was here for part of the day, 1 out of 5 adults total). Last night I saw we were trending toward 30 kWh and ended up with 28. That’s not far above our goal (good), but it is far above the past several days, especially for being a “quiet” day (bad).

So, I took a look at the power meter over the past few days.

Looks like what I’m calling the “average floor” (if you squint, the solid horizontal line where most of the electrical activity seems to settle) rose sometime Saturday evening. Interesting. Looking closer, it looks like a 300-500W increase between 7 and 8 pm on Saturday evening.

It would be nice if the Sense app could take the guesswork out of this and just show me a horizontal line at this floor and label it (or give me the number somewhere).

Now, I just need to figure out what turned on Saturday night to cause this increase… today we are trending to exceed our goal again.


#11

A renter with a smaller heater that knew you would t be home to monitor?