How nguyenthiphuongchi05/ttnt works
This repository is a diabetes-risk prediction web app backed by a separately trained Random Forest model. The principal workflow is user-entered health measurements → validation and preprocessing → probability-based risk classification and explanation → rendered result. A training workflow reads the Pima dataset, prepares features, compares and tunes models, selects a recall-oriented threshold, and saves artifacts for inference. The training excerpts omit some intermediate code; links below follow the visible main workflow and documented preprocessing behavior.
Main entry points: app.py Flask prediction handler · train.py Training workflow · utils.py Artifact persistence · preprocessing.py Feature preprocessing
Components 16
Model training
- Diabetes dataset
diabetes_prediction/data/diabetes.csv - Training workflow
diabetes_prediction/train.py - Feature preprocessing
diabetes_prediction/src/preprocessing.py - Model evaluation
diabetes_prediction/train.py - Training reports
diabetes_prediction/reports
Risk inference
- Flask prediction handler
diabetes_prediction/app.py - Input validation
diabetes_prediction/app.py - Sample transformation
diabetes_prediction/src/preprocessing.py - Risk and factors
diabetes_prediction/app.py
Web interface
- Result page
diabetes_prediction/templates/index.html - Browser interactions
diabetes_prediction/static/script.js - Page styling
diabetes_prediction/static/style.css
Model artifacts
- Trained model
diabetes_prediction/model/diabetes_model.pkl - Preprocessing artifacts
diabetes_prediction/model/artifacts.pkl - Artifact persistence
diabetes_prediction/src/utils.py
Other components
- User
Connections 17
- User → Flask prediction handler: submits measurements
diabetes_prediction/app.py - Diabetes dataset → Training workflow: provides data
diabetes_prediction/train.py - Training workflow → Feature preprocessing: fits transforms
diabetes_prediction/train.py - Training workflow → Feature preprocessing: transforms test data
diabetes_prediction/train.py - Training workflow → Model evaluation: compares models
diabetes_prediction/train.py - Training workflow → Training reports: writes metrics
diabetes_prediction/train.py - Training workflow → Artifact persistence: saves artifacts
diabetes_prediction/train.py - Artifact persistence → Trained model: writes model
diabetes_prediction/src/utils.py - Artifact persistence → Preprocessing artifacts: writes transforms
diabetes_prediction/src/utils.py - Flask prediction handler → Artifact persistence: loads artifacts
diabetes_prediction/app.py - Flask prediction handler → Input validation: validates fields
diabetes_prediction/app.py - Flask prediction handler → Sample transformation: transforms sample
diabetes_prediction/app.py - Flask prediction handler → Trained model: predicts probability
diabetes_prediction/app.py - Flask prediction handler → Risk and factors: classifies and explains
diabetes_prediction/app.py - Flask prediction handler → Result page: renders result
diabetes_prediction/app.py - Browser interactions → Flask prediction handler: submits form
- Page styling → Result page: styles page