Hi,
It seems you are trying to compare the output value from the sensor against reference data, observing both the temperature and original analogue signal to make the necessary adjustment.
One approach can be collecting reference data under controlled conditions, where you know the true concentration of the substance you're detecting, as well as the corresponding sensor output and temperature. Use the data to train a model that predicts the true concentration from the raw sensor output and temperature. Depending on how temperature, sensor readings, and actual concentration are related, you can use machine learning methods like linear or polynomial regression or more advanced ones like neural networks.
Refer to the below documentation for more information on the linear regression function "fitlm", and "predict" function:
