apply for Learnable calls only discrete value.

[kordjamshidi]

Spotted by @onlinecco, when the regressor is called regressor(x) an NPE is thrown. For now the solution is that we should call it by saying explicitly regressor.realValue(x).

[danyaljj]

What is a "regressor"?

[kordjamshidi]

A Saul Learnable that does regression and returns a real value instead of class label.

[danyaljj]

Do we have examples for regression?

[kordjamshidi]

i added the biological example to Joey's fork. It is in the WebApp RP.

[danyaljj]

Just for future reference, on regression Learners (which I don't know how many they are), realValue should be called, instead of discreteValue: example

[kordjamshidi]

did you add any information here to what I said above?!

[danyaljj]

I added the reference to the LBJava function which needs to be called, if someone wants to fix it in future.

[kordjamshidi]

I already said we should call realValue and then we do not need to do anything with Lbjava we should just do a kind of type check/conversion in Saul when writing apply for Learnable.

[danyaljj]

Good. Tnx.

[kordjamshidi]

@onlinecco do you remember that we could train a regressor with multiple real properties? it seems it does not work now :(. We need a simple test for regression.