PyRecs 0.0.2

Recommender systems for Python

A Python Recommender Systems library

Installation

I have only tested on Python 3.5.1

pip3 install pyrecs

Usage

Loading data from a Pandas DataFrame:

>>> import io
>>> import pandas as pd
>>> from pyrecs import collab
>>>
>>> df = pd.read_csv(io.StringIO("""
...        ,Torchy's,Tacodeli,In-N-Out,P. Terry's,Casa de Luz,Koriente
... Sam,           5,        ,       4,         4,          3,       1
... Matthew,        ,       2,       1,          ,          5,       5
... Sarah,         5,       4,       2,         2,          5,       5
... Hannah,         ,        ,       1,         1,          5,
... """.replace(' ', '')), index_col=0)
>>>
>>> cf = collab.CollaborativeFiltering()
>>> cf.fit(df)
>>> print(cf.predict([('Sam', 'Tacodeli'), ('Hannah', 'Koriente')]))
[ 3.41666667  5.76851363]

Predicting ratings based on training data:

>>> import numpy as np
>>> import pyrecs
>>> from sklearn.cross_validation import train_test_split
>>>
>>> data = [[10,     3.4, np.nan, None],
...         [10,     0,   10,     5],
...         [np.nan, 1.4, 10,     3],
...         [np.nan, 8,   2,      5]]
>>>
>>> X, y = pyrecs.collab.matrix_to_dataset(data)
>>> X_train, X_test, y_train, y_test = train_test_split(X, y)
>>>
>>> cf = pyrecs.collab.CollaborativeFiltering()
>>> cf.fit(X_train, y_train)
>>>
>>> cf.predict(X_test)
array([ 0.25,  3.4 ,  9.75])
>>> y_test
[1.4, 10, 8]

Tests

py.test tests