Dataset Masking: Per-Sample Training Weight

[alexjc]

Closes #125.

[alexjc]

This apparently breaks the code with non-masked data. Looking into it.

[paullo0106]

@alexjc hi Alex, I just pip install scikit-neuralnetwork (0.4) and would like to use nn.fit(X_train, y_train, w_train) as shown in the Per-sample weighting example. However, I don't see that w variable in my fit() function in the sknn source code therefore I couldn't pass in the parameter. What could be the reason? Thank you.

[alexjc]

We haven't yet pushed fit() with weight to PIP. It's in master only.

There's a different documentation page for stable (linked from PIP) here: scikit-neuralnetwork.readthedocs.org/en/stable/guide_beginners.html

[paullo0106]

Thanks Alex, let me take a look!