Compare three AI approaches: Naïve Bayes, Genetic Algorithm (weights), and GA Feature Selection
CSC425 AI Term Project by Zack Sargent | Slides
Type your own review or click an example below:
Analyzing with all three models...
Description: Uses Bayesian Networks (pgmpy) with conditional independence assumption
Strengths: Fast, interpretable, best overall accuracy (81%)
Weaknesses: Cannot capture word dependencies like 'not good'
Test Accuracy: 81.00%
Training Time: 0.41s
Description: Genetic Algorithm evolving 501 weight values for classification
Strengths: Demonstrates evolutionary search in high-dimensional space
Weaknesses: Poor performance (54.5%) - shows GA limitations as classifier
Test Accuracy: 54.50%
Training Time: 0.62s
Description: GA selects optimal features, then trains Naïve Bayes on them
Strengths: Best recall (86.27%), demonstrates GA as optimization tool
Weaknesses: Very slow training (23 minutes), slight accuracy drop vs pure NB
Test Accuracy: 77.50%
Training Time: 1407s
Explore the internal state of each model to understand how they work