Implementation of ID3, in MATLAB.
No builds necessary, just open MATLAB interpreter in the root folder of this project. But the data to train the tree must have an specific format... Basically:
- Import the data to train the tree into a numerical matrix.
- Make sure the "class" to classify must be on the first column of the matrix of data, so the rest of the columns should be features.
To generate the tree just create a new object id3
data = csvread('fisher.csv')
example = id3
example.createTree(data)
example.plot()
The Tree is generated, the find method to test the model in to be implemented... soon.