Dr. Hanan Shteingart
Dr. Hanan Shteingart
suggest solution in PR #4
yes, the evaluation should prevent time leakages - training should be based on "past" and testing on "future". training should not have access to "future" data. "past" and "future" definitions...
It's pretty straight forward to add it to the python wrapper. Am I mistaken? I can contribute if needed. On Wed, Dec 5, 2018, 19:46 annaveronika We want the algorithm...
changing the source code to ``` # set the treatment column to zero (the control group) X_new = np.hstack((np.zeros((X.shape[0], 1)), X))**.copy()** ``` solved the issue.
Hi, This is my first contribution to a GitHub public project, so I am excited :-) I find that `FeatureUnion` is superior because everything is in terms of Scikit-Learn native...
I feel that if we move to feature union we should actually PR scikit learn so much larger community can enjoy it. What do you say? On Sep 3, 2016...
Asked here: https://github.com/scikit-learn/scikit-learn/issues/7334
the data format is explained in the readme.txt file. Each line should have a count of number of tokens followed by a term index and it counts. The term index...
I also notice that some of the preprocessing steps of moving from `ap.txt` to `ap.dat` are missing. What stop words were removed? What tokenization processed was used. This stuff is...
please pull request the necessary changes so we can all enjoy the correction.