ExaScience / smurff

Bayesian Factorization with Side Information in C++ with Python wrapper
MIT License
70 stars 14 forks source link

Errors in example (documentation) #117

Closed jaak-s closed 5 years ago

jaak-s commented 6 years ago

Few examples are incorrect on this page: https://smurff.readthedocs.io/en/latest/notebooks/different_noise_models.html

The examples of "Adaptive noise", "Binary matrices" get in Y parameter, which is not supported, should be Ytrain. Also aux_data should be removed, it is not supported anymore, right? The line numbers in the text of "Binary matrices" are off:

Care has to be taken to make input to the model, as operating with sparse matrices can drop real 0 measurements. In the below example, we first copy the matrix (line 9) and then threshold the data to binary (line 10).

Also "Tensor Factorization" example has aux_data argument.

tvandera commented 6 years ago

Thanks for reporting this. Indeed, those examples are quite outdated. I'll fix them. Tom

tvandera commented 5 years ago

Docs fixed:

https://smurff.readthedocs.io/en/latest/notebooks/different_noise_models.html

Please use smurff-0.14, as 0.13.x contains a bug in the ProbitNoise.