Orange3 Recommendation

Orange3 Recommendation is a Python module that extends Orange3 to include support for recommender systems. All features can be combined with powerful data mining techniques from the Orange data mining framework.

Dependencies

Orange3-Recommendation is tested to work under Python 3.

The required dependencies to build the software are Numpy >= 1.9.0 and Scikit-Learn >= 0.16

Install

This package uses distutils, which is the default way of installing python modules. To install in your home directory, use::

python setup.py install --user

To install for all users on Unix/Linux::

python setup.py build
sudo python setup.py install

For development mode use::

python setup.py develop

Widget usage

After the installation, the widgets from this add-on are registered with Orange. To run Orange from the terminal, use

python3 -m Orange.canvas

new widgets are in the toolbox bar under Recommendation section.

Scripting

All modules can be found inside orangecontrib.recommendation.*. Thus, to import all modules we can type:

from orangecontrib.recommendation import *

Rating pairs (user, item):

Let's presume that we want to load a dataset, train it and predict its first three pairs of (id_user, id_item)

>>> import Orange
>>> from orangecontrib.recommendation import BRISMFLearner
>>> data = Orange.data.Table('movielens100k.tab')
>>> learner = BRISMFLearner(K=15, steps=25, alpha=0.07, beta=0.1)
>>> recommender = learner(data)
>>> prediction = recommender(data[:3])
>>> print(prediction)
[ 3.79505151  3.75096513  1.293013 ]

Recommend items for set of users:

Now we want to get all the predictions (all items) for a set of users:

>>> import numpy as np
>>> indices_users = np.array([4, 12, 36])
>>> prediction = recommender.predict_items(indices_users)
>>> print(prediction)
[[ 1.34743879  4.61513578  3.90757263 ...,  3.03535099  4.08221699 4.26139511]
 [ 1.16652757  4.5516808   3.9867497  ...,  2.94690548  3.67274108 4.1868596 ]
 [ 2.74395768  4.04859096  4.04553826 ...,  3.22923456  3.69682699 4.95043435]]

Performance

MovieLens100K: (Loading time: 0.428s)

Algorithm	RMSE	MAE	Train time	Settings
Global Average	1.126	-	0.001s	-
Item Average	1.000	-	0.001s	-
User Average	1.031	-	0.001s	-
User-Item Baseline	0.938	-	0.001s	-
BRISMF	0.823	-	1.453s/iter	k=15; alpha=0.07; beta=0.1
SVD++	0.836	-	7.991s/iter	k=15; alpha=0.07; beta=0.1

MovieLens1M: (Loading time: 4.535s)

Algorithm	RMSE	MAE	Train time	Settings
Global Average	1.117	-	0.010s	-
Item Average	0.975	-	0.018s	-
User Average	1.028	-	0.021s	-
User-Item Baseline	0.924	-	0.027s	-
BRISMF	0.872	-	14.347s/iter	k=15; alpha=0.07; beta=0.1
SVD++	0.896	-	114.665s/iter	k=15; alpha=0.07; beta=0.1

MovieLens10M: (Loading time: 49.804s)

Algorithm	RMSE	MAE	Train time	Settings
Global Average	1.060	-	0.129s	-
Item Average	0.942	-	0.256s	-
User Average	0.970	-	0.256s	-
User-Item Baseline	0.877	-	0.361s	-
BRISMF	0.841	-	138.309s/iter	k=15; alpha=0.07; beta=0.1

Relevant links

• Official source code repo: https://github.com/salvacarrion/orange3-recommendation
• HTML documentation: http://orange3-recommendation.readthedocs.io
• Download releases: https://github.com/salvacarrion/orange3-recommendation/releases
• Issue tracker: https://github.com/salvacarrion/orange3-recommendation/issues