search

ragulpr / wtte-rnn

WTTE-RNN a framework for churn and time to event prediction
MIT License
762 stars 186 forks source link

Minor adjust in shape of return #29

Closed sn3fru closed 6 years ago

sn3fru commented 6 years ago

When we pass an 'x' shape to 'y' without the setting, count nan will return globally (a single number) instead of returning an array with the count of each 'x' counts.

ragulpr commented 6 years ago

Hi, thank you for contributing and sorry for not going into depths on this before.

The reason for using np array instead of looping and building a regular array of integers is (very marginal) speed gains but primarily that the resulting seq_lengths-array is used here and there for numpy-operations such as seq_lengths.sum() or subsetting.

What is the purpose of this approach?