Graphing Sensor Values with linear interpolation instead of staircase

[GrahamS (Graham Smith)]

I have a sensor which stores the high and low tide heights over time.
I would like join the points with a line like you see in this Excel chart.

Home Assistant sensor and input_number present the points as a staircase with a horizontal line followed by a vertical line between points.
I looked into Mini Graph Chart, but the problem seems to originate with the HA database format. I would have to create more points between high and low tides to get a sensible graph.
(I had a similar issue back Dec 2020. At that time, the solution was to use input_number. After HA revisions both sensor and input plot as staircases.)

[Ildar_Gabdullin]

@GrahamS

but the problem seems to originate with the HA database format.

Mini graph card takes data from DB in accordance with points_per_hour, then calculates extra points if needed.
Stepline seems to show really acquired data. Direct lines shows assumed values.

[GrahamS (Graham Smith)]

This is an old issue that seems to have been missed. See

A guide to smoother plots using input_number

I recently posted a compliant that sensor and input number were plotted differently on the Lovelace History Graph Card. I turned this difference into a plotting benefit for my DYI weather station sensors. (The weather station reports temperature, pressure, humidity and battery voltage every 5 minutes but only significant changes. The hardware draws an average of 20 uA from the battery due to these update constraints.) The following pictures show the plot differences: [outdoor_temperature] …

Maybe there needs to be an option for the sensor state_class attribute so that measurement points are connected with linear interpolation when plotted. This would also result in smoother plots without the need for Mini Graph Chart or oversampling.

[Ildar_Gabdullin]

@GrahamS

This would also result in smoother plots without the need for Mini Graph Chart or oversampling.

“Smoother plots” and w/o “oversampling”?
It is not possible. To draw a curve, you need a set of data. These data either may be “real” (taken from a physical object) or “assumed” (“oversampling”).

[GrahamS (Graham Smith)]

By oversampling, I mean creating many more data points so that the small staircase steps are not as visible. If you use linear interpolation, ie draw a line between data points then you will get a smoother looking graph without using as many points.
Balloob thinks that linear interpolation is creating “fake” data. See his comments here:
https://www.reddit.com/r/homeassistant/comments/1axjebo/smooth_graph_lines/?rdt=60628
The “real” data is just the value, time pairs when measured. HA is using rectilinear interpolation when graphing and is creating a “fake” data point, holding the last known value at the time of the next value change.
Moving on to Apex Charts but not hopeful.

[Ildar_Gabdullin]

@GrahamS

Balloob thinks that linear interpolation is creating “fake” data

Agree. These are fake assumed data.

[GrahamS (Graham Smith)]

Good News!
Apexcharts has interpolation options that produce the desired graphical effect.
Here is the test code example:

type: custom:apexcharts-card
header:
  show: false
  title: Tide Chart
  show_states: true
  colorize_states: true
apex_config:
  stroke:
    width: 2
    curve: straight
  markers:
    size: 5
graph_span: 100h
series:
  - entity: sensor.temperature_humidity_xs_sensor_humidity

Here is the graph using the straight option:

Here is the graph using the stepline option:

Here is the HA history graph:

The HA history graph matches the Apexchart graph with the stepline option.
Note that the orange dots are the real points from the HA database.
This just shows two methods of interpolation between real points.
Notice that there are a lot of corners on the stepline graph that don’t have a orange dot.
In my opinion, the straight option would be a useful addition to the HA history graph card.
It is wrong to say that one interpolation method is more “fake” than another.

[Ildar_Gabdullin]

Here I see a stepline

and understand that between 2 points there are no other readings.

Here I see a line

and can make a false assumption about a value in a particular point between these 2 known points.

It is up to a particular person how to be deceived.