Bool sensor predict / probability of kid turning on PC?

My kid has some automated blinds that I want to avoid activating excessively.

I have setup a WOL binary sensor that already detects when my kid turns on his PC.

eg. of excessive use of blinds.

  1. It’s morning, the PIR sensor detects movement => blinds are opened as per automation.
  2. 5 min. later my kid turns on his PC, the blinds are closing as per automation.

Ideally I want to predict the probability of my kid turning on his PC before he actually does it. If the probability is high then do not bother to open blinds when the morning automation runs.

Is there a “magic” binary sensor platform that I can use for tracking this using historic data of the PC WOL sensor itself ?

It should work based on the actual time (datetime now). I should be able to ask the sensor : “is his PC mostly on during this very specific point in time ?”

Then we have weekends and vacation to further complicate matters.

I know about the bayesian sensor platform but am unsure if this is the right platform for my use case.

The closest we have to a predictive sensor is this:

Weighted existing sensor states (not past) are used to predict a likelihood.

1 Like