csrank 1.2.1

Context-sensitive ranking

CS-Rank

CS-Rank is a Python package for context-sensitive ranking and choice algorithms.

We implement the following new object ranking/choice architectures:

• FATE (First aggregate then evaluate)

• FETA (First evaluate then aggregate)

In addition, we also implement these algorithms for choice functions:

• RankNetChoiceFunction

• GeneralizedLinearModel

• PairwiseSVMChoiceFunction

These are the state-of-the-art approaches implemented for the discrete choice setting:

• GeneralizedNestedLogitModel

• MixedLogitModel

• NestedLogitModel

• PairedCombinatorialLogit

• RankNetDiscreteChoiceFunction

• PairwiseSVMDiscreteChoiceFunction

Check out our interactive notebooks to quickly find out what our package can do.

Getting started

As a simple “Hello World!”-example we will try to learn the Pareto problem:

import csrank as cs
from csrank import ChoiceDatasetGenerator
gen = ChoiceDatasetGenerator(dataset_type='pareto',
                                n_objects=30,
                                n_features=2)
X_train, Y_train, X_test, Y_test = gen.get_single_train_test_split()

All our learning algorithms are implemented using the scikit-learn estimator API. Fitting our FATENet architecture is as simple as calling the fit method:

fate = cs.FATEChoiceFunction()
fate.fit(X_train, Y_train)

Predictions can then be obtained using:

fate.predict(X_test)

Installation

The latest release version of CS-Rank can be installed from Github as follows:

pip install git+https://github.com/kiudee/cs-ranking.git

Another option is to clone the repository and install CS-Rank using:

python setup.py install

Dependencies

CS-Rank depends on Tensorflow, Keras, NumPy, SciPy, matplotlib, scikit-learn, scikit-optimize, joblib and tqdm. For data processing and generation you will also need PyGMO, H5Py and pandas.

Citing CS-Rank

You can cite our arXiv papers:

@article{csrank2019,
  author    = {Karlson Pfannschmidt and
               Pritha Gupta and
               Eyke H{\"{u}}llermeier},
  title     = {Learning Choice Functions: Concepts and Architectures },
  journal   = {CoRR},
  volume    = {abs/1901.10860},
  year      = {2019}
}

@article{csrank2018,
  author    = {Karlson Pfannschmidt and
               Pritha Gupta and
               Eyke H{\"{u}}llermeier},
  title     = {Deep architectures for learning context-dependent ranking functions},
  journal   = {CoRR},
  volume    = {abs/1803.05796},
  year      = {2018}
}

License

Apache License, Version 2.0

History

Unreleased

No changes yet.

1.2.1 (2020-06-08)

• Make all our optional dependencies mandatory to work around a bug in our optional imports code. Without this, an exception is raised on import. A proper fix will follow.

1.2.0 (2020-06-05)

• Change public interface of the learners to be more in line with the scikit-learn interface (ongoing). As part of these changes, it is no longer required to explicitly pass the data dimensionality to the learners on initialization.

• Rewrite and document normalized discounted cumulative gain (ndcg) metric to fix numerical issues. See #32 for details.

• Fix passing fit keyword arguments on to the core network in FATEChoiceFunction.

• Fix arguments for AllPositive baseline.

• Raise ValueError rather than silently using a default value for unknown passed arguments.

• Internal efforts to increase code quality and make use of linting (black, flake8, doc8).

• Remove old experimental code.

1.1.0 (2020-03-19)

• Add the expected reciprocal rank (ERR) metric.

• Fix bug in callbacks causing the wrong learning rate schedule to be applied.

• Make csrank easier to install by making some dependencies optional.

• Add guidelines for how to contribute to the project.

1.0.2 (2020-02-12)

• Fix deployment to GH-pages

1.0.1 (2020-02-03)

• Add HISTORY.rst file to track changes over time

• Set up travis-ci for deployment to PyPi

1.0.0 (2018-03-05)

• Initial release