How to predict using multiply equations?

[leelew]

Hi,
Thanks for the useful tools! I really enjoy using it.

I want to test the performance of result equations on test dataset. But model.predict() only return the predictions from best model. I found it can pass index into predict, but I cannot find index of all candidate equations. Is there any methods to predict by all equations shown in hall of frame file?

Regard,
Lu

[MilesCranmer]

Thanks!

The equations are stored in model.equations_ in a pandas dataframe.

For example:

model  # fitted PySRRegressor

for i in range(len(model.equations_)):
    print("Equation", i, "equal to", model.sympy(i))
    print("with predictions")
    pred = model.predict(X, i)
    print(pred)

This assumes 1-dimensional y. For 2 dimensional you would need to iterate over each component too.

[leelew]

Thanks for your useful reply! It works well!