How javencp/diabetes_readmission_project works
This repository is a diabetes 30-day readmission prediction project combining SQLite data preparation with Python machine learning. Its main workflow builds a database and feature table from included hospital encounter CSVs, loads and cleans those features, splits by patient to reduce leakage, encodes categorical variables, fits three classifiers, and evaluates them. Separate notebooks document exploratory analysis, tuning, and model comparison; SQL analysis is another product capability. SQL relationships are based on the README, since those files were not sampled.
Main entry points: run_pipeline.py Pipeline Runner · db SQLite Database · schema.sql Database Schema · load_data.sql Load and Clean
Components 15
SQL Data Preparation
- Encounter CSVs
data/diabetic_data.csv - Database Schema
sql/schema.sql - Load and Clean
sql/load_data.sql - Feature Builder
sql/features.sql - SQLite Database
db
ML Pipeline
- Pipeline Runner
python/run_pipeline.py - Feature Loader
python/src/data_loader.py - Clean and Encode
python/src/preprocessing.py - Patient Split
python/src/split.py - Three Classifiers
python/run_pipeline.py - Metric Evaluation
python/src/evaluate.py - Saved Models
python/run_pipeline.py
Analysis and Tuning
- SQL Analysis
sql/data_analysis.sql - EDA and Tuning
python/notebooks
Other components
- Project User
Connections 18
- Project User → Database Schema: creates schema
README.md - Encounter CSVs → Load and Clean: imports records
README.md - Load and Clean → SQLite Database: loads cleaned data
README.md - Database Schema → SQLite Database: defines tables
README.md - SQLite Database → Feature Builder: supplies records
README.md - Feature Builder → SQLite Database: builds features
README.md - SQLite Database → Feature Loader: reads feature table
python/src/data_loader.py - Pipeline Runner → Feature Loader: loads features
python/run_pipeline.py - Pipeline Runner → Clean and Encode: cleans and encodes
python/run_pipeline.py - Pipeline Runner → Patient Split: splits and selects
python/run_pipeline.py - Pipeline Runner → Three Classifiers: fits classifiers
python/run_pipeline.py - Pipeline Runner → Metric Evaluation: computes metrics
python/run_pipeline.py - Pipeline Runner → Saved Models: saves models
python/run_pipeline.py - Project User → Pipeline Runner: runs pipeline
README.md - EDA and Tuning → Three Classifiers: tunes model families
README.md - Project User → EDA and Tuning: explores and compares
README.md - Project User → SQL Analysis: runs analysis
- SQL Analysis → SQLite Database: queries