Utils

recoder.utils.normalize(x, axis=None)

Returns the normalization of x along axis.

Parameters:

• x (np.array) – matrix or vector
• axis (int, optional) – the axis along which to compute the normalization

recoder.utils.dataframe_to_csr_matrix(dataframe, user_col, item_col, inter_col, item_id_map=None, user_id_map=None)

Converts a pandas.DataFrame of users and items interactions into a scipy.sparse.csr_matrix.

This function returns a tuple of the interactions sparse matrix, a dict that maps from original item ids in the dataframe to the 0-based item ids, and similarly a dict that maps from original user ids in the dataframe to the 0-based user ids.

Parameters:

• dataframe (pandas.DataFrame) – A dataframe containing users and items interactions
• user_col (str) – users column name
• item_col (str) – items column name
• inter_col (str) – user-item interaction value column name
• item_id_map (dict, optional) – A dictionary mapping from original item ids into 0-based item ids. If not given, the map will be generated using the items column in the dataframe
• user_id_map (dict, optional) – A dictionary mapping from original user ids into 0-based user ids. If not given, the map will be generated using the users column in the dataframe

Returns:

A tuple of the csr_matrix, a dict item_id_map, and a dict user_id_map

Return type:

tuple