Goal: Study and compare various supervised classification techniques.
For each classifier, the same workflow was followed:
- Data preprocessing (standardization)
- Splitting the dataset into training and testing sets
- Model training
- Performance evaluation (confusion matrix & classification accuracy)
- Parameter tuning (especially optimizing the K value, where applicable)
- Gaussian Naive Bayes: Provides a probabilistic approach assuming feature independence.
- Support Vector Classification (SVC): Constructs an optimal hyperplane to separate classes.
- Decision Tree Classifier: Applies "if-then" rules in a hierarchical tree structure.
- Gradient Boosting Classifier: Sequentially builds trees where each one corrects the errors of the previous.
The results highlighted that each algorithm has specific strengths and weaknesses depending on the dataset characteristics.
Hyperparameter optimization played a important role in improving model accuracy. In particular, graphical analysis of error rates as a function of different K values helped identify the optimal configurations for each algorithm.