sgadgil6 / cs4641-unsupervised-learning Goto Github PK

The code can be found at https://github.com/sgadgil6/CS4641-Unsupervised-Learning


WEKA was used to perform the dimensionality reduction and python's scikit-learn, a machine learning library for Python, was used to run the neural network learner. I used a repository to obtain the code for running the neural network. The GitHub link to the repo is  https://github.com/qliuan/CS4641-Machine-Learning/tree/master/3/qliu359/repo

Installation:

WEKA
1. Install Java 8 
2. Install WEKA from https://www.cs.waikato.ac.nz/ml/weka/downloading.html
3. Run the algorithms using WEKA

Python scripts
This project requires numpy and sklearn to be installed.

Dataset:

The datasets used are located in the /data directory. The MNIST dataset is stored in a csv file called MNIST_data.csv. The Breast Cancer Dataset is stored in a file called breastCancer_data.csv

Folder Structure:

The main folder contains all the python files and the shell scripts needed for running the code. The /data directory contains the original datasets, the clustered datasets for breast cancer, the arff files generated by WEKA, and the excel files used for analysis. 

File Structure:

In the /data directory the datasets obtained by dimensionality reduction are named as breastCancer_algorithm.csv. For example breastCancer_ica.csv referes to the projected data produced by ICA. For the clustered datasets, file names used are breastCancer_algorithm_clu_new.csv

Running the python scripts:

All the code can be run by using the shell scripts provided in the main directory. To run a shell script, open a terminal and type ./<script_name>.sh

run_clu_nn.sh script runs the neural network learner on the dataset with clustering features. run_nn.sh script runs the neural network on the dimensionally reduced dataset.