Thanks, i did History stats sensor
sensor:
- platform: history_stats
name: Pump ON today
entity_id: binary_sensor.nasos
unique_id: pumpon.nasos
state: "on"
type: count
start: "{{ today_at() }}"
end: "{{ now() }}"
But i can’ show cycles per hour, in statistics card “stat types” i can’t choose state or change option… So i see only increasing values.
How can i visualise only per hour changes ?
This works for me:
- platform: history_stats
name: Pump On 1H Count
unique_id: pump_on_1h_count
entity_id: binary_sensor.pump_cycle_on
state: "on"
type: count
end: "{{ now() }}"
duration:
hours: 1
As Capt. Tom showed above, adjust your History Stats duration to be an hour, not a day as you have it now.
I’ve changed to 1 hour, and it was what i needed at first look, but why it shows like this ?
History from 3am till 4am - 9 time “on”
History from 4am till 5 am - 0 time “on”
But statistics bar show like this - 9 times each hour (from 3 till 4, and from 4 till 5)
Where i’m doing wrong again ?
Yeah, that doesn’t seem right. Maybe post your YAML here. Usually when I get a result like that it’s something stupid I did, like a typo in the event ID, or just looking at the wrong entity. Those are the hardest to find. Sometimes going away for a while and coming back later with a fresh outlook makes it jump out at me.
Here’s sensor section
sensor:
- platform: history_stats
name: Pump ON today
entity_id: binary_sensor.nasos
unique_id: pumpon.nasos
state: "on"
type: count
end: "{{ now() }}"
duration:
hours: 1
Here’s another example of strange counting
Stats graph says that from 9 till 12 - pump was on 1 time in each hour.
However history shows that from 10 till 11 it was never on
12 till 13 - correct both places
13 till 14 - totally different
Btw here’s similar problem - Statistics graph card showing wrong data
What attributes are you using in the statistics graph?
Should be “change”.
Or try using the history graph instead of the statistics graph.
Unfortunately “change” or “state” options unavailable from stats card configuration - it’s greyed out.
That’s why i’m stucked)
And History Graph is totally different - it doesn’t represent “per hour cycles”.
Maybe it’s just a bug ?
Ok you can only use change with state_class: total (or total_increasing). But forget that.
Looking again at your graph it is working fine. From 3am to 4am it turned on 9 times. And that is what your graph shows at 4am. Remember you defined it as now minus one hour:
end: "{{ now() }}"
duration:
hours: 1
So the 4am bar is for 3am to 4am.
For 4am to 5am the count was 0. And that is what shows for the 5am bar, at least I think it does it is hard to see behind the tool tip.
There is no bug. It is working as expected.
Ok, minus 1 hour, i got it. So here’s another recent examples.
23pm bar (should be 22-23 right?) - 4 times
History bar 22-23 - 1 time
0 bar (should be 23-0 right?) - 6 times
History bar 23-0 - 4 times
1am bar (should be 0-1am right?) - 3 times
History bar 0-1am - 3 times
It’s somewhere right, somewhere wrong… And i can’t find logic here
P.S. Changed from end: “{{ now() }}” to start: “{{ now() }}” to see what happens…
Perhaps it is because it is a sliding one hour window ending now not “on the hour”. What time did you make those graphs?
Um if you make it start now with a duration of one hour that probably requires time travel to collect data. Now to one hour in the future?
To get it to reset on the hour you will have to do something like this which will collect all day starting at 0:00:
start: "{{ today_at() }}"
end: "{{ now() }}"
Then feed that to an hourly cycle utility meter. Like you were doing, but now with history stats sensor, not stats sensor.
ooohhh… i think i got the main idea…
But can’t we just say something like “beginning of each hour” with some templating ? and duration 1 hour ?
Or why it’s impossible to make smth like this ?
start: "{{ today_at() }}"
end: "{{ now() }}"
duration: 1 hour
The thing is to get almost realtime stats…
I think this might do what you want:
end: "{{ now().replace(minute=0,second=0,microsecond=0) }}"
duration: 1 hour
Thanks, i’ll give a try!
I just updated the template. Both should work but the new version is simpler (only one replace instead of three).
If you replace end with start in Tom’s template you’ll get closer to real-time updates. The counter should increase within a minute of a state change, and it should reset to zero or 1 (see below) within a minute after each new hour.
A peculiarity of the history_stats count method is that it is NOT a “rising edge” counter. Instead it counts the number of times the device has been in the desired state during that time window. In other words, if you are counting on states, and the device was already on at the start of the window, the counter will count that as 1, even though there is no rising edge. That is important because if the pump is already running when a new hour starts, you’ll get a count of 1 added in the previous hour and a count of 1 added in the new hour, even though the pump only turned on once.
If you want a rising edge counter that updates immediately upon state changes and returns to zero immediately at the start of every hour, then the only option is a trigger-based template sensor with some fancy templates.
If the peculiarities of the history_stats don’t bother you, then there’s no need to go down that path.
To be honest i’ve started this “investigaion” after i’ve made up garden water output and started changing my water pressure tank and relay settings for maximum capacity (minimising water pump starts/stops).
So +/- 1 start/stop isn’t pain in the ass…
But last template sems to works strange as well…
If you want to do the template method, you can use this:
template:
- trigger:
- platform: state
entity_id: binary_sensor.nasos
to: 'on'
- platform: time_pattern
hours: '/1'
sensor:
- name: Water Pump Hourly Counter
unique_id: 6759e0f7-e104-4690-a700-3fce4d92bddc
state_class: measurement
state: "{{ 0 if trigger.platform == 'time_pattern' else this.state | int(0) + 1 }}"
It will be unknown until it gets triggered for the first time.
Seems it works more mystique than before:)
History - water softener rinse from 3 till 4 am - 9 times, and 3 time from 8 till 9 am
Stats
Is this the correct sensor in your history stats?
It’s named Hacoc in your graph.
