AI-weka

Event Type Prediction with WEKA

Overview

This project implements an AI-based event type prediction system as part of the 7051CEM Web Applications and AI module at Coventry University. The goal is to predict the type of an event (Conference, Wedding, Workshop, Party) based on its attributes (location, date, description) using the WEKA API. This is Part B of the "Event Management Platform" coursework, focusing on machine learning integration.

Objective

Train a classification model to predict event types using a dataset of event listings.
Use WEKA’s J48 decision tree classifier for prediction.
Evaluate the model with 10-fold cross-validation and predict new events.

Dataset

The dataset (events.arff) contains 80 rows, with a balanced distribution across four event types (20 rows each for Conference, Wedding, Workshop, Party). Attributes include:

Location: Nominal (13 UK cities: London, Manchester, Coventry, etc.).
Date: String, converted to nominal using StringToNominal.
Description: String, processed with StringToWordVector to extract word counts.
EventType: Target class (Conference, Wedding, Workshop, Party).

Implementation

Classifier: WEKA’s J48 decision tree.
Preprocessing:
- StringToNominal: Converts the date attribute to nominal.
- StringToWordVector: Converts the description attribute into word count features.
Training: The model is trained on the dataset using 10-fold cross-validation.
Prediction: The trained model predicts the event type for new instances.

Technologies Used

Java: Core programming language.
WEKA API: For machine learning (J48 classifier).
Maven: Dependency management (WEKA library).

Steps

Clone the Repository:

git https://github.coventry.ac.uk/guggillan/AI-weka
cd EventPrediction

Step 2:
```
mvn clean install
```

Step 2:

mvn exec:java -Dexec.mainClass="org.eventclassifier.classifier.App"

Results

The J48 classifier was evaluated using 10-fold cross-validation, achieving the following results:

Accuracy: 82.05% (64/78 instances correctly classified).
Kappa Statistic: 0.7591
Mean Absolute Error: 0.1046
Root Mean Squared Error: 0.2793

Detailed Accuracy by Class

Class TP Rate FP Rate Precision Recall F-Measure MCC ROC Area PRC Area Conference 0.857 0.105 0.750 0.857 0.800 0.723 0.931 0.761 Wedding 1.000 0.000 1.000 1.000 1.000 1.000 1.000 1.000 Workshop 0.810 0.140 0.680 0.810 0.739 0.636 0.876 0.700 Party 0.611 0.000 1.000 0.611 0.759 0.740 0.862 0.783 Weighted Avg 0.821 0.066 0.847 0.821 0.820 0.767 0.916 0.805

Confusion Matrix

a b c d <-- classified as 18 0 3 0 | a = Conference 0 18 0 0 | b = Wedding 4 0 17 0 | c = Workshop 2 0 5 11 | d = Party

License

guggillan/AI-weka

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Contributors

Uh oh!

Languages