Checking the Relevance of the Inputs

During data import, you can check the inputs' relevance to the outputs (see also To check the relevance of the inputs).

If you do so, a polynomial stepwise regression is done with the inputs and outputs. The stepwise regression ignores inputs with a significance < 5% and can find dependent inputs.

If, for example, the training data contains the inputs speed, load and speed + load, then one of the inputs has a low significance.

The order of the inputs is important. After the stepwise regression, the inputs are permuted column by column, and a pseudo RMSE is calculated per input, to get a heuristic of the input's relevance. The findings are then plotted in the "Relevance of Inputs" window.