How bhupendra-jangir/vip works
Vehicle Insurance Prediction is an end-to-end training and serving application. Users submit vehicle attributes through the web app for a prediction, or trigger retraining; retraining ingests MongoDB data, validates and transforms it, trains and evaluates a model, then pushes an accepted model to S3. The app’s prediction pipeline loads the production model from S3. Training orchestration and the data-access/storage layers are shown; several component internals were not sampled, so uncertain internal calls are omitted.
Main entry points: training_pipeline.py Training pipeline · app.py Web routes · aws_storage.py S3 storage client · prediction_pipeline.py Prediction pipeline
Components 17
Web application
- Web routes
app.py - Prediction result
app.py
Model training
- Training pipeline
src/pipline/training_pipeline.py - Data ingestion
src/components/data_ingestion.py - Data validation
src/components/data_validation.py - Data transformation
src/components/data_transformation.py - Model trainer
src/components/model_trainer.py - Model evaluation
src/components/model_evaluation.py - Model pusher
src/components/model_pusher.py
Data access
- Mongo data access
src/data_access/vip_data.py - Mongo connection
src/configuration/mongo_db_connection.py - S3 storage client
src/cloud_storage/aws_storage.py
Model serving
- Prediction pipeline
src/pipline/prediction_pipeline.py - S3 estimator
src/entity/s3_estimator.py
External systems
- MongoDB Atlas
- S3 model registry
Other components
- User
Connections 19
- User → Web routes: submits vehicle data
- Web routes → Training pipeline: starts training
- Training pipeline → Data ingestion: invokes
- Training pipeline → Data validation: invokes
- Training pipeline → Data transformation: invokes
- Training pipeline → Model trainer: invokes
- Training pipeline → Model evaluation: invokes
- Training pipeline → Model pusher: invokes if accepted
- Data ingestion → Mongo data access: exports records
- Mongo data access → Mongo connection: gets connection
- Mongo connection → MongoDB Atlas: connects to
- Model evaluation → S3 storage client: loads production model
- Model pusher → S3 storage client: uploads accepted model
- S3 storage client → S3 model registry: reads and writes
- Web routes → Prediction pipeline: requests prediction
- Prediction pipeline → S3 estimator: predicts with
- S3 estimator → S3 storage client: loads model
- Prediction pipeline → Web routes: returns prediction
- Web routes → Prediction result: renders result