The sensor used in the A3 is either the BME680 or the BME280 from Bosch. These are high spec sensors, I would expect very good linearity.

Maybe not the sensor, but the influence of the case and fun may be not linear depending of the temperature?

From you last post the correction would be T:-3.8 and H:+8.83

Looking the humidity, now that we have less I have the impression that the difference is smaller, therefore I would point to a non linear correction.
You know far better than me the sensor would it be possible that it is not linear?

Have a look of what I noticed

Temperature, Humidity and CO2:

The Oregon Scientific is at the same location of the A3, the KNX Siemens not exactly at the same position.
For this measures I left the door and the windows closed to reduce the air moving influencing them.

uRADMonitor A3: 25.6 °C, 55.17 %RH CO2 386.79 ppm
Oregon Scientific: 21.8 °C, 64 %RH
KNX Siemens AQR: 21.2 °C, 67.24 %RH CO2 562.88 ppm

Regarding CO2 I can not tell who is better.
For the temperature (maybe higher because inside of the case) and humidity (maybe due to the fun blowing air and higher temperature).

Noise:

When I measure for example with my mobile phone near the A3 31dB and 1meter from it I have max 23dB.
And the noise may for sure be influenced by the fun…or ?

VOC AIQ:

Device 820000FA at 14:23 September the 4th of 2018.
The VOC AIQ reports 348 and this is treated as severely polluted, this device is currently on my office at home (50cm from me).

Now (10.09.2018 10:42) for the date time 14:23 September the 4th of 2018.
The VOC AIQ reports something like 17 and not anymore 348??
Hos is this calculated? all values at each request? When inserting the data?

PS: most of this is coming from two entries in the community, but I have the impresison that the support is the best place and I have maybe better chance to get an “official” response.

WOuld it be possible to have the possibility to inject all the measures in a InfluxDb ?