How sanskritisonee/shipment-sure works
Shipment Sure is an end-to-end shipment-delay prediction project, from data preparation and model training to an interactive Streamlit application. Planners enter shipment details or upload a CSV; the app loads the trained model and feature schema, predicts delay risk, and can show SHAP explanations. The README documents the upstream cleaning, EDA, and training stages, but their notebook implementations were not sampled, so those links reflect documented workflow rather than inspected calls.
Main entry points: app.py Shipment UI · shipment_delay_model.pkl Delay model · modeltraining.ipynb Model training · datasets Anomaly cleaning
Components 15
Data preparation
- Data understanding
dataexploration.ipynb - Anomaly cleaning
datasets - EDA and features
eda.ipynb
Model training
- Model training
modeltraining.ipynb
Prediction application
- Shipment UI
app.py - Feature assembly
app.py - Single prediction
app.py - Risk classification
app.py - SHAP explanation
app.py - CSV batch input
app.py - Batch prediction
app.py - Prediction results
app.py
Model artifacts
- Delay model
shipment_delay_model.pkl - Feature schema
model_features.pkl
Other components
- Planner
Connections 19
- Planner → Shipment UI: enters shipment
README.md - Planner → CSV batch input: uploads CSV
README.md - Shipment UI → Feature assembly: collects details
app.py - Shipment UI → Delay model: loads model
app.py - Shipment UI → Feature schema: loads schema
app.py - Feature assembly → Single prediction: builds input
app.py - Single prediction → Delay model: predicts with
app.py - Single prediction → Risk classification: classifies probability
app.py - Single prediction → SHAP explanation: explains with
app.py - CSV batch input → Batch prediction: validates CSV
app.py - Batch prediction → Delay model: predicts with
app.py - Batch prediction → Prediction results: adds predictions
app.py - Risk classification → Prediction results: displays risk
app.py - SHAP explanation → Prediction results: displays impacts
app.py - Data understanding → Anomaly cleaning: informs cleaning
README.md - Anomaly cleaning → EDA and features: provides clean data
README.md - EDA and features → Model training: provides features
README.md - Model training → Delay model: produces model
README.md - Model training → Feature schema: produces schema
README.md