Thanks for that @HomeAgain, I had not heard of ConstantGraph.
In theory it looks nice. But it doesnât look worth it to me.
By now I have 7 years of data, every 10 minutes, for about 60 sensors. Looks like ConstantGraph canât touch that, not even with itâs highest paid tier. (Summarizing âDaily data for 5 yearsâ would only allow for once a day, not every 10 minutes, and itâs not clear how many sensors it can handle.)
Their homepage looks wonky and practically broken on Windows Chrome, which is a red flag relative to how smooth it is to use and how well and long itâll be supported.
I wonder why these companies donât hook into an individualâs storage space, like the CShwer log did with Google Sheets. It would make things much less expensive for them, and users would own and be responsible for their own data. If this makes deep analysis slower (because it would take many minutes to analyze a million datapoints fetching external data), I guess I can live with that as the price one has to pay.
When I set up my big smart home network in 2016 (about two dozen devices and 100 sensors in all), I had dreams of being able to chart, e.g., temperatures and humidity inside and outside my house relative to HVAC usage and the weather in general. So I could answer questions like, How much energy do I use depending on the weather in summer and winter. How much does a cloudy day versus a sunny day matter?
But in retrospect, this was incredibly naive at many levels. No consumer-grade smart home software comes anywhere close to doing this. In fact, all smart home apps are practically made for children. They have big fat displays for single values at one point in time (right now), instead of dense tables of values for numerous devices that can be graphed over time. Add onto that a never ending stream of various ways that even their very basic functionality doesnât always work. Everything from very poor notifications relative to battery levels, to what it does when the hub or internet or even power goes out, plus add in the various ways that hubs update (and things go out for a while until you fix something) plus changing security and changing back ends over time (like, the loss of CSchwer functionality now). Also I havenât seen a single one that includes a way to estimate values across the many data gaps in time that are definitely going to happen. No smart home company even has the faintest mention (like in their ads) of how they address data gaps, when they really should if theyâre serious about data collection. They just arenât.
I used to think SmartThings was the smart personâs choice, because they let the community develop apps to supplement it. But now I see that this is more like how many companies have fobbed off actual depth and/or support to community forums, where you have to search for and ask questions, instead of actually being able to talk to someone responsible and knowledgable at the company itself. Dev support is nice, but if the core product itself isnât great or well supported - and keeps changing, usually just sideways in ways that donât really improve it but do cause headaches - well, there you go.
In Samsungâs defense, their model of selling a hub once (relatively cheaply) and getting no more income per user, just isnât sustainable for producing an extremely deep and wide end-user experience. It just isnât. Nor is anything short of a relatively expensive smart home subscription, that I imagine few would go for. Really the model here is more like (chemical) industrial facilities that have very deep process monitoring. But of course, theyâre astronomically too expensive for simple end users.
It also seems to me that one would need something more like AI to successfully take thousands or millions of an individualâs datapoints (if weâre talking multiple years, multiple sensors, every 10 minutes or whatever) and combining it into something useful and wholistic. It would have to be smart enough to handle gaps in data as well as other vagaries, like moving a particular sensor from time to time. Even down to things like how, every time I replace the battery on an outdoor sensor, it briefly has indoor readings when it comes back up. Or comparing a particular temperature sensor against a temperature standard, and adjusting it relative to that. (Another digression: Why donât any of these sensor makers hardly ever even mention whether it used a standard when manufactured, and supply detailed info on how sensitive and accurate they are?) There are many such vagaries in data taken over long periods which have to be addressed to have an accurate overall picture.
Okay, this is all TL;DR, laugh. Iâll stop here.