COMP9321-Project

Introduction

In this assignment, we created a web app based on a heart diease dataset and use machine learning algorithms to make predictions on important factors relevant to heart dieases.

Organisation

Firstly, we divide our project into two part: backend and frontend. Secondly, we use Machine Learning - decision tree algorithm to make prediction. Then we use cross validation method to increase the accuracy of the prediction.

Tasks

• Dataset Collection
• UI Design
• Machine Learning
• Optimization algorithm
• API Server

Frame

• Frontend
• Backend
• README.md

RawData

Dataset from http://www.cse.unsw.edu.au/~cs9321/19T1/assn/heart.tar.

DataCleaning

• Convert data to csv format
• Drop Invalid Data
• Fill Missing Data By Median or Average

ModelsTrainning

• Building decision tree model to make prediction
• As for decision tree classifier model,we used the information gain and entropy split criteria. We also calculated the accuracy of our decision tree model.

Optimization Algorithm

• Use Cross-validation method to improve accuracy of prediction

Clustering

• Use K-means and PCA method to cluster features

Tech Stacks

• Frontend : Vanilla Javascript, JQuery, HTML, Materialize CSS
• Backend : Python (Flask), pandas, SQLite
• Machine Learning : Scikit-Learn

Repo

COMP9321 Ass3

Installation & How to Run

• Backend

$ cd backend
$ pip install -r requirements.txt
$ python3 app.py

WARNING: Backend URL must be at http://127.0.0.1:5000

• Frontend

$ npm install -g http-server
$ cd frontend
$ http-server

Open browser at http://localhost:8080

Instructions for reproducing prediction model

• Simply run predict.py and the model will be automatically generated