A Decision tree inducer (originally based on ID3).
- Numeric features
- Categoric features
- Asynchronous branch induction (for performance)
$ ./tr -h Decision tree inducer Version (of 2012-10-27): 0.1.1 All rights reserved Isak Karlsson 2012+ Example: tr -i <EXAMPLES> -y 0.66 -d 5 tr -i <EXAMPLES> -y 1 -d 0 -y [Percent training data] A percentage of data that is used for training the model. The rest is used for validation. Default: 0.66 -d [Maximum depth to paralellize] The depth in which the model is induced in paralell. Default: 5 -ag  Calculate gain in parallel even thought the MAX_DEPTH is reached (-d). -o  Output the tree model -p [N instances required to partition] Stop inducing a branch (and take the majority) when N is reached (for a given branch) Default: 10 For example, $ git clone .... $ cd erlang-id3/src $ ./tr -i ../data/bank.txt -y 0.66 -d 10 -ag -p 100 -o > model.txt Running: * File: "../data/bank.txt" * Split: 0.66 * Depth: 10 * Prune: 100 * Gain async: true Took: 1.159709 Accuracy: 0.8792531390280398