The librec-auto project aims to automate recommender system experimens using LibRec.
Project description
============ Librec-auto
About
librec-auto
is a Python tool for running recommender systems experiments.
It is built on top of the open-source LibRec_ package, and
can take advantage of the many algorithm and metric implementations there.
.. _LibRec: https://github.com/guoguibing/librec
The basic element of librec-auto
execution is the "study", which is a series
of experiments carried with a single algorithm, a single data set, and a set
of evaluation metrics. The experiments differ from each other by the hyperparameters
given to the algorithm. librec-auto
allows such studies to be conducted with
minimal experimenter intervention, and supports such capabilities as:
- recommendation result re-ranking
- re-running evaluations without re-computing results
- summarizing results in output graphs
- integration with Slack and Dropbox
Workflow
The workflow of an study involves identifying appropriate data, creating training / test splits, implementing or choosing algorithms, running experiments (possibly with a range of different parameters), and reporting on the results.
Configuration
Librec-auto uses an XML-based configuration system similar to Maven or Ant.
Project structure
This directory contains the Python libraries for the librec_auto module. There are two other affiliated respositories:
librec-auto-java
_: Contains the java source for the wrapper between LibRec and librec-auto, which is implemented in theauto.jar
file.librec-auto-sample
_: Contains sample data and configuration files that can be used to explore the functionality of librec-auto
.. _librec-auto-java: https://github.com/that-recsys-lab/librec-auto-java .. _librec-auto-sample: https://github.com/that-recsys-lab/librec-auto-sample
Repo structure
/jar
: Contains the jar files for LibRec and the wrapper./rules
: Contains the rules for translating configuration data to LibRec properties format./librec_auto
: Contains the Python code for the project./doc
: Contains documentation for the project/test
: Contains the unit tests (not many right now)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for librec_auto-0.1.50-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 1400e25e8f9176f8ff9b552ab03f0050f5b7692d013f35a77fb763432b55199c |
|
MD5 | 298e975e3e26a59a6a26fe9f2163e8d0 |
|
BLAKE2b-256 | 03ad4a41eb9363104de757e2d47ee165a9c4d219fb32b389682973ff5eaa2de7 |