Heart Disease Classification Analysis

Data Intake & Approval. Datasets can be selected from the bundled regional sources or uploaded as CSV files. Normal uploads are mapped to the bundled schema, checked for missing, invalid, and out-of-range values, and approved with explicit review decisions before they become available for training. Trusted inputs can use the same canonical schema without the review step. Exploratory distributions, correlations, and custom plots provide additional context for each approved dataset.

Training & Evaluation. Three models are fully implemented for the current analysis: Naive Bayes, K-Nearest Neighbors (KNN), and Support Vector Machine (SVM). Additional models are planned as the project develops. Each training run uses the columns and rows approved during Add Data intake with its configured missing-value strategy, then reports accuracy, precision, recall, and F1 score across stratified validation folds. Saved results can be compared across datasets and model runs before using a selected model for an individual prediction.