rfgb

Relational Functional Gradient Boosting in Python.


Keywords
relational-learning, statistical-relational-learning
License
GPL-3.0
Install
pip install rfgb==0.3.1

Documentation

Relational Functional Gradient Boosting (rfgb)

Python Package Index (PyPi) latest version. License. Master branch build status. Master branch code coverage. Documentation build status and link to documentation.

rfgb: Relational Functional Gradient Boosting (in Python).

Installation

Stable builds may be installed from PyPi

pip install rfgb

Or develop further by cloning the repository

git clone https://github.com/hayesall/rfgb.git
cd rfgb/
python setup.py develop

Quick-Start

Learning with a relational dependency network.

cd testDomains/Logistics/
rfgb init
rfgb learn rdn -target unload

Additional options for each subcommand may be viewed by passing -h as a parameter.

rfgb -h
rfgb learn rdn -h

Classification with Expert Advice (-advice)

Preferred and non-preferred labels may be provided as advice during classification via logical rules. This advice may be specified in a file named advice.txt in the train directory for a dataset.

Four datasets (BlocksWorld, HeartAttack, Logistics, and MoodDisorder) have an advice file included for demonstration

  1. Logistics
cd testDomains/Logistics/
rfgb init
rfgb learn rdn -advice -target unload
  1. HeartAttack
cd testDomains/HeartAttack/
rfgb init
rfgb learn rdn -advice -target ha

Targets

"Targets" specify what is learned, examples of the target are provided in pos.txt, neg.txt, or examples.txt (for regression). These are specified here for convenience.

Dataset Target
BlocksWorld putdown
BostonHousing medv
HeartAttack ha
Insurance value
Logistics unload
MoodDisorder bipolar
TicTacToe put or dontput
ToyCancer cancer
XOR xor

In Development

  • Test cases (codecov >90%)
  • General interaction improvements for commandline and library
  • Learning Markov Logic Networks

License

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

A full copy of the license is available in the base of this repository. For more information, see https://www.gnu.org/licenses/