AI Survey Program

AI/ML Political Affiliation Prediction System

Written in: Java
Author: Aaron Ballesteros
Course: Computer Science 311 - Artificial Intelligence
Date: March 15, 2023

Table of Contents

• About
• Features
• Dependencies
• Implementation Details
• Visuals
• Version History

About

Data mining is pivotal in understanding customer/member behaviors. This Java program leverages machine learning to predict a user's political leaning based on a survey. The core objective is for the program to guess a user's political party before they complete the survey, enhancing user experience and data accuracy.

Features

• Dynamic Survey System: Presents questions with answer options varying based on political beliefs.
• Data Collection & Storage: Collects and stores user responses efficiently.
• Advanced Prediction: Uses a trained machine learning model for accurate political affiliation prediction.
• Data Visualization: Visual representations of the collected data for better insights.

Dependencies

The program harnesses the power of the weka library for machine learning and data classification.

import weka.classifiers.Classifier;
import weka.classifiers.trees.J48;
import weka.core.Attribute;
import weka.core.DenseInstance;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.converters.ConverterUtils.DataSource;

Implementation Details

Data Collection

The program likely incorporates standard Java I/O methods to interactively gather user input. After collecting responses, they might be stored in separate CSV files corresponding to each political party.

public void collectUserData() {
    Scanner scanner = new Scanner(System.in);
    System.out.println("Enter your response: ");
    String response = scanner.nextLine();
    // Additional logic to process and store the response
    appendToCSV(determinePartyFile(response), response);
}

Machine Learning Model

With the help of the weka library, a classifier (e.g., J48) is trained on the collected data to build a prediction model.

public void trainClassifier() {
    // Load data
    Instances trainingData = new Instances(new BufferedReader(new FileReader("data.csv")));
    trainingData.setClassIndex(trainingData.numAttributes() - 1);

    // Build classifier
    Classifier classifier = new J48();
    classifier.buildClassifier(trainingData);
}

Prediction Mechanism

Once the model is trained, it can be used to predict a user's political affiliation based on their survey responses.

public String predictAffiliation(String userInput) {
    // Convert userInput into an Instance format
    Instance userInstance = ...;

    // Predict
    double predictedClass = classifier.classifyInstance(userInstance);
    String predictedAffiliation = trainingData.classAttribute().value((int) predictedClass);

    return predictedAffiliation;
}

Data Handling

The program will have functions dedicated to reading and writing data, especially for interacting with CSV files.

public void readFromCSV(String fileName) {
    // Logic to read from CSV
    ...
}

public void appendToCSV(String fileName, String data) {
    // Logic to append data to CSV
    ...
}

Visuals

Program Preview

Data Visualization

Version History

• Version 1 (3/20/23): Initial Commit.
• Version 2 (3/22/23): UI Design, Dataset Data addition, Datawaste removal, Bonus question, Code cleanup.
• Version 3 (3/22/23): Data storages, CSV Reading/Writing enhancements.