pattern_project
CECS 550-01 | Project 1 - DTrees Ensemble
Team Name: Classy Fires
Team Members: Sophanna Ek, Melissa Hazlewood Rowan Herbert, Dennis La
Introduction: The automated Ensemble Decision Tree (DTree) Builder was built by using a weighted vote of the two DTrees based on their accuracy. The first DTree was built with an initial selection of the vectors and the second DTree was built with boosted feature vectors. The same data was used in both DTrees for training, testing and validating. The DTree builder used the standard Shannon-based Entropy and Information Gain mechanisms with a Training Set. The boosted feature vectors were built by the union of the first Training set and Holdout Set boosted with two extra mis-classified Holdout vectors. This project implemented in Python.
File Contents:
- Dtree.py: Decision tree class includes all neccessary functions to construct the decision tree.
- DtreeNodes: Decision tree data structure includes question and leaf nodes.
- Ensemble.py: Decision tree Ensemble class include all neccessary function to construct the decision tree Ensemble.
- parse.py: Data input processing file.
- second_tree.py: Decision tree #2 construction.
- main.py: The entry of the application. It prints out all the output for this project.
- data_set.csv: data resource file
External Requirements:
- None
Setup and Installation:
- Download and install Python intepreter
--
pip install python - Unzip the project folder
Sample Invocation:
python3 main.py- or
python main.py > output.txt
Features:
- Boosted Feature Vectors
- Weighted Voting Using Accuracy Rate
Issues:
- None so far