Project Title: Classifiers Machine Learning
Datasets: https://archive.ics.uci.edu/dataset/891/cdc+diabetes+health+indicators
Project Title: Health and Medicine Data Classification
Associated Tasks: Classification
Instances: 253,680
Subject Area: Health and Medicine
Feature Type: Categorical, Integer
Features: 21
Data Cleaning and Preprocessing: Feature Selection: - Selected relevant features such as BMI, blood pressure, and dietary habits based on domain knowledge and initial correlation analysis.
Normalization:
- Applied Min-Max scaling to normalize numerical variables like age and cholesterol levels.
Models Explored: - K-Nearest Neighbors (KNN) - Random Forest - Gradient Boosting - Adaptive Boosting - Bagging Classifier - Rationale for Model Selection: - The Bagging Classifier was selected due to its superior accuracy compared to other models, achieving an accuracy score of 84%.
Metrics: Evaluated model performance using: - Accuracy - Precision - Recall
Cross-Validation:
- Conducted cross-validation to assess model robustness and generalizability.
License: - https://creativecommons.org/publicdomain/zero/1.0/