COMP4121_Project

This repo contains all the code and the test data used in my comp 4121 project on implementations of classification algorithms. I implemented 4 models, K-Nearest Neighbours, Decision Tree and Naive Bayes.

Dependencies

This program will work with any version of Python 3.X and uses Pandas, Numpy and Sklearn. (However it does not use the SKlean models.)

Running the Algorithms

The program has an entrypoint at ./run.py. The top of the file is shown in the snippet below;

if __name__ == '__main__':
    runner = LogisticRegression()
    runner.run(dataset=Dataset.Cancer)

    # runner = KNN()
    # runner.run(dataset=Dataset.Cancer)

    # runner = NaiveBayes()
    # runner.run(dataset=Dataset.Cancer)

    # runner = DecisionTree()
    # runner.run(dataset=Dataset.Cancer)

Simply comment and uncomment different models, and run them with Python run.py.

Fitting to other datasets.

In the Models/Helpers.py file, it is possible to import other datasets that are added to the directory. A Enum called StdDistributionType provides two options, Labeled or RealValued. For each feature in the dataset, label it one of these features and and split it ms.test_train_split(...). Use the cancerData() method as a template.