Struggling with incorrect min / max / distance_absolute values

Configuration
1

I might be overlooking something obvious, but I just cannot understand this issue I am experiencing with the statistics integration.

I have a simple counter, which tracks the number of heating cycles of my heat pump.

grafik
grafik705×615 23.9 KB

I now want to derive the number of heating cycles over the past 24h. This should be a simple max - min calculation, ie 7424 - 7417 = 7. But HA gives me 6!?

I have tried statistics change, I have tried statistics distance_absolute, both give me 6. (See my sensor configuration below).

I then looked at min and max values, and the underlying issue seems to be that the min value returned is 7418, rather than 7417.

I also tried adding keep_last_sample: true (I have to admit, from the docs I still do not full understand what it does), but that made no difference.

Can anybody help me wrap my head around this?

- platform: statistics
  name: "WP Heizzyklen 24h"
  entity_id: sensor.cycles_waermepumpe
  unique_id: heizzyklen_waermepumpe_flamme
  state_characteristic: change
  max_age:
    hours: 24
  precision: 0
- platform: statistics
  name: "WP Heizzyklen 24h (Distance)"
  entity_id: sensor.cycles_waermepumpe
  unique_id: heizzyklen_waermepumpe_flamme_2
  state_characteristic: distance_absolute
  max_age:
    hours: 24
  precision: 0
- platform: statistics
  name: "WP Heizzyklen 24h (Max)"
  entity_id: sensor.cycles_waermepumpe
  unique_id: heizzyklen_waermepumpe_flamme_max
  state_characteristic: value_max
  keep_last_sample: true
  max_age:
    hours: 24
  precision: 0
- platform: statistics
  name: "WP Heizzyklen 24h (Min)"
  entity_id: sensor.cycles_waermepumpe
  unique_id: heizzyklen_waermepumpe_flamme_min
  state_characteristic: value_min
  keep_last_sample: true
  max_age:
    hours: 24
  precision: 0
2

The arrow in the graph below points to the first state change in the last 24 hours. This is important because the sensor only updates on state changes:

c617b0679284801b049ecc77144dd5d2acf2c30f
c617b0679284801b049ecc77144dd5d2acf2c30f705×615 24.1 KB

This is why you are seeing 7418 as the minimum. After the first state change in the 24 hour period the value was 7418.

3

Thank you for the quick response! That is what I had feared (and is really confusing behavior, I wonder what happens if there is no increase in the counter over 24h).

So is there any straightforward way to get what I am looking for? The numerical increase in my counter over 24h? That must be something users would require frequently. Right away, my rain counter would come to mind (which I now realize has also occassionally been giving me slightly off values, probably for the same reason).

As a last resort, I can create this via pyscript and running an SQL query against the short-term statistics, as I already did to measure counter increase for the current day. But I would think there must be a simpler solution?