decision-tree-algorithms

Implementation of some decision tree algorithms in Python.

Requirements

Python 3+

All dependencies defined in requirements.txt. Dependencies can be installed by:

pip install -r requirements.txt

ID3 Algorithm

ID3 (Iterative Dichotomiser 3) is a classification algorithm, proposed by Ross Quinlan in 1986, that follows a greedy approach of building a decision tree by selecting the best attribute that yields the maximum Information Gain. Only symbolic variables are processed, numerical ones must be discretized to be considered.

How to use:

Run the ID3/main.py file with the following arguments:

python ID3/main.py [data file] [target variable] [instance to classify (optional)]

• data file: .csv or .xlsx file containing a categorical dataset with a target variable.
• target variable: name of the variable (or column) holding the classification label.
• instance to classify: .json file with a dictionary containing the values of a new object (or instance) to be classified from the newly generated decision tree.

Output

The program returns a new decision tree, in a textual format. For example, the following decision tree:

is represented in textual format as:

[F1:c->(F)b->(T)a->[F2:y->(F)x->(T)]]

If the user provides a new instance to be classified, the program also returns the classification label for that instance and the path through the decision tree that the algorithm used to classify it.