Daniil Potapov

Is ItemItemRecommender inconsistent?

4

Hi, thanks for this nice package! While working with the library I came across the following inconsistent behavior of [**ItemItemRecommender**](https://github.com/benfred/implicit/blob/master/implicit/nearest_neighbours.py#L12): 1. **ItemItemRecommender** can generate less than N recommendations (while expected...

ValueError: Input contains NaN, infinity or a value too large for dtype('float64').

4

Code: ```python ... train['title'].isnull().sum() # Out: 0 title_countvec = CountVectorizer(ngram_range=(1,3), max_features=300000, lowercase=True) title_countvec.fit(train['title'], y_train) train_title_countvec = title_countvec.transform(train['title']) title_vectorizer = TfIcfVectorizer(norm='l2', sublinear_tf=True) title_vectorizer.fit(train_title_countvec, y_train) train_title_countvec = title_countvec.transform(train['title']) np.isfinite(train_title_countvec.data).all(), np.isinf(train_title_countvec.data).any() # Out:...

Code: ```python from lightning.classification import AdaGradClassifier ``` Traceback: ```python ImportError Traceback (most recent call last) in () 17 from sklearn.model_selection import GridSearchCV 18 ---> 19 from lightning.classification import AdaGradClassifier 20...

Windows + python=3.9 is not supported

2

I can't install rectools on Windows for python=3.9 due to problems with installing [nmslib](https://github.com/nmslib/nmslib) (one of the dependencies). Issue for tracking - https://github.com/nmslib/nmslib/issues/464

If `user_id` and `item_id` columns are CategoryDType, then np.setdiff1d works very slowly on large volumes (>10 million unique ones) Possible solution is to replace: https://github.com/MobileTeleSystems/RecTools/blob/76c41e0e039cd050b46ec0f6cb7f0f668fca9574/rectools/model_selection/time_split.py#L146 with ```python new_users = set(df_test[Columns.User].unique())...