I’d like to mention that with the new ESPHome release (1.13.6) there’s very little need to calibrate the sensors. Apparently, there was a data-type mismatch (https://github.com/esphome/esphome/pull/532) which was causing the poor correlation from X/Y/Z and Temp sensors and now the data looks pretty darn good:
I set my input_selects using all accel and gyro values and it’s is really representative of the actual washer/dryer state.