clyamar

Simple forest learning algorithm for Clojure.

Installation

1. Make sure you have Leiningen installed

2. Clone project from GitHub.

   git clone https://github.com/xhresko/clyamar.git

3. Create uberjar for standalone usage.

   cd clyamar

   lein uberjar

Usage

Run the app with args

$ java -jar clyamar-0.2.0-standalone.jar [options] action

Options

--input-file - file with samples (in LightSVM format)

--model-file - file for model saving/loading

--depth - tree depth

--granularity - number of thresholds used for feature while creating a split

--alpha - learning rate

--trees - number of trees in the forest

Actions

train - create model on given labeled dataset

eval - evaluate given model on given labeled dataset

predict - print predictions for given dataset

Examples

Download dataset for regression task from Eunite 2001 competition.

$ wget https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/regression/eunite2001 -O /tmp/eunite2001.libsvm

Start learning using Leiningen (while inside the project directory).

$ lein run train -i /tmp/eunite2001.libsvm

Bugs

• problems with some LightSVM data format "implementations"
• different results of MSE for evaluation and training

License

Copyright © 2016

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.