Decision Tree Classifier

Introdution

This is a repository for a university project.

The subject is "Metodi statistici per l'apprendimento" and we want to do a tree classifier.

Thanks to @random-forests for the tutorial on developing a decision tree classifier from scratch. Part of scripts are made by him in a youtube tutorial.

You can see in the history what has be done by me.

Dependencies

Most of the modules are built in, but you can install the remaining ones with: pip install x

These are the modules used in this project:

• csv
• math
• numpy
• pathlib
• time
• os.path
• matplotlib
• argparse

Dataset setup

1. put the dataset data (.csv) in "Datasets"
2. put the dataset header (.csv) in "Datasets". Just the titles of the columns

How to use the scripts

(These instructions are for windows)

1. Go in the Source folder and use '''python DecisionTreeClassifier.py'''. If you want to see the resulting tree, use -d at the end: '''python DecisionTreeClassifier.py -d'''
2. Write the relative Path inside the "Datasets" folder for the dataset data. For example, to use the dataset 'car_data.csv' inside the 'CarEvaluation' folder, we'll write "CarEvaluation\car_data.csv".
3. Write the relative Path inside the "Datasets" folder for the header of the dataset. For example, to use the header 'car_header.csv' inside the 'CarEvaluation' folder, we'll write "CarEvaluation\header_data.csv".
4. Write the number of the indixes that have a numeric value. This is to distinguish the cases '<=' to '=='. In case there are no numeric columns, just use the 'Enter' key.

If it doesn't do exception, good job, you just need to wait :D

How to use WEKA

1. Open Weka (assuming it is installed)
2. Go to "simple CLI"
3. Use the instruction in the "Weka Instruction" file. Note: the path should be the absolute file of the dataset. Use the .arff file. The data are the same, but there are written also the attributes.