How yashuu05/bike-rental-prediction works
Souel Bike Rental Prediction is an end-to-end bike-demand ML project with a FastAPI prediction service and an offline training workflow. The main user path is browser input to API, preprocessing, model inference, and optional database persistence; training selects and tunes a model for later inference. The README also describes ETL, validation, DVC data versioning, experiment tracking, AWS storage, and Airflow orchestration. Some of those capabilities are documented but their implementation links are not shown in the sampled source, so those links are omitted or marked unverified.
Main entry points: main.py FastAPI routes · pipeline.py Training pipeline · ingestion.py Data ingestion · predict.py Prediction pipeline
Components 27
Prediction service
- Web interface
app/static/index.html - Browser behavior
app/static/js/main.js - FastAPI routes
app/main.py - Request schemas
app/schemas.py - Prediction records
app/models.py - MySQL database
app/database.py
ML lifecycle
- Prediction pipeline
src/prediction/predict.py - Saved preprocessor
models/preprocessor.joblib.dvc - Trained model
models/model.joblib.dvc - Training pipeline
src/training/pipeline.py - Base model training
src/training/train_1.py - Hyperparameter tuning
src/training/train_2.py - Model evaluation
src/model_evalution.py - Model configuration
configs/models.py
Data preparation
- Feature engineering
src/features/feature_engg.py - Dataset extraction
src/extraction/extract.py - Data ingestion
src/ingestion/ingestion.py - Data validation
src/validation/validate.py - Preprocessing
src/preprocessing/transform.py
Data and operations
- Artifact paths
configs/paths.py - File utilities
utils/utilities.py - S3 integration
src/aws/s3.py - Airflow DAG
dags/exampledag.py - Versioned datasets
data/raw/input_features.csv.dvc
Other components
- Bike renter
- Public dataset
- AWS S3
Connections 28
- Bike renter → Web interface: enters inputs
README.md - Web interface → Browser behavior: loads scripts
- Browser behavior → FastAPI routes: submits request
- FastAPI routes → Request schemas: validates request
app/main.py - FastAPI routes → Prediction pipeline: loads resources
app/main.py - FastAPI routes → Prediction pipeline: preprocesses input
app/main.py - FastAPI routes → Trained model: predicts with
app/main.py - FastAPI routes → Saved preprocessor: transforms with
app/main.py - Prediction pipeline → Feature engineering: creates features
src/prediction/predict.py - FastAPI routes → Prediction records: creates record
app/main.py - FastAPI routes → MySQL database: commits prediction
app/main.py - FastAPI routes → MySQL database: queries history
app/main.py - Public dataset → Dataset extraction: provides dataset
- Dataset extraction → Data ingestion: feeds ingestion
- Data ingestion → Data validation: passes data
- Data validation → Preprocessing: validated data
- Preprocessing → Feature engineering: engineers features
src/preprocessing/transform.py - Training pipeline → Base model training: selects model
src/training/pipeline.py - Training pipeline → Hyperparameter tuning: tunes selection
src/training/pipeline.py - Training pipeline → Model configuration: uses models
src/training/pipeline.py - Base model training → Model evaluation: evaluates models
src/training/train_1.py - Hyperparameter tuning → Model evaluation: evaluates tuned model
src/training/train_2.py - Training pipeline → File utilities: saves model
src/training/pipeline.py - Training pipeline → Artifact paths: uses artifact path
src/training/pipeline.py - Data ingestion → S3 integration: uses S3 helper
src/ingestion/ingestion.py - S3 integration → AWS S3: stores data
- Public dataset → Versioned datasets: tracks versions
README.md - Airflow DAG → Dataset extraction: orchestrates ETL