There are a number of existing issues with my bayesian binary_sensor sleep tracker. Works well 95% of the time, but need to make it perfect.
Currently the battery state going to "Charging" or "Full" has too much weight. It triggered me as sleeping when I plugged the phone into the car and the state went to full, even though I was away from home.
Need to add more criteria that I'm actually home. Check if binary_sensor.sean_presence is off or binary_sensor.car_presence is on, either of those things mean I'm not home and can't be sleeping.
The bedtime routine automation has conditions in place to prevent it from triggering under these situations, but I need to add similar probability values to the sensor itself. Because the "Time Slept" sensor gets info directly from the state of binary_sensor.sleep and it'll be inaccurate otherwise.
There are a number of existing issues with my bayesian binary_sensor sleep tracker. Works well 95% of the time, but need to make it perfect.
Currently the battery state going to "Charging" or "Full" has too much weight. It triggered me as sleeping when I plugged the phone into the car and the state went to full, even though I was away from home.
Need to add more criteria that I'm actually home. Check if binary_sensor.sean_presence is off or binary_sensor.car_presence is on, either of those things mean I'm not home and can't be sleeping.
The bedtime routine automation has conditions in place to prevent it from triggering under these situations, but I need to add similar probability values to the sensor itself. Because the "Time Slept" sensor gets info directly from the state of
binary_sensor.sleep
and it'll be inaccurate otherwise.