How ferozewaris/fraud-credit-card-transaction-detection works
This repository implements a PySpark fraud-detection workflow: it loads transaction data, cleans and engineers features, trains and compares classification models, and saves transformed test results. Its documented product is a Spark ML pipeline intended to run on Spark clusters. The source also includes time-feature transforms and a deployment pipeline, but deployment is a stub and time featurization is not shown in the sampled workflow. Pipeline entry points and external invocation are not documented, so those connections are omitted.
Main entry points: training.py Training pipeline · feature.py Preprocessing pipeline · database.py Data file I/O · base.py Pipeline interface
Components 15
Data preparation
- Preprocessing pipeline
fraud_transaction_detection/pipeline/feature.py - Data transformers
fraud_transaction_detection/data_processing/transformers.py
Feature engineering
- Feature pipeline
fraud_transaction_detection/pipeline/feature.py - Feature transforms
fraud_transaction_detection/features/transformers.py - Time featurizer
fraud_transaction_detection/features/transformers.py - Time strategies
fraud_transaction_detection/features/constants.py
Model training
- Training pipeline
fraud_transaction_detection/pipeline/training.py - Model tuning
fraud_transaction_detection/pipeline/training.py - Model selection
fraud_transaction_detection/pipeline/training.py - Training results
fraud_transaction_detection/result_engine/training.py - Result interface
fraud_transaction_detection/result_engine/base.py
Pipeline runtime
- Data file I/O
fraud_transaction_detection/utils/database.py - Pipeline interface
fraud_transaction_detection/pipeline/base.py - Deployment pipeline
fraud_transaction_detection/pipeline/deployment.py
Other components
- Pipeline initiator
Connections 18
- Pipeline initiator → Preprocessing pipeline: invokes
- Preprocessing pipeline → Data transformers: configures
fraud_transaction_detection/pipeline/feature.py - Preprocessing pipeline → Data file I/O: loads data
fraud_transaction_detection/pipeline/feature.py - Preprocessing pipeline → Feature pipeline: builds and runs
fraud_transaction_detection/pipeline/feature.py - Feature pipeline → Data transformers: configures
fraud_transaction_detection/pipeline/feature.py - Feature pipeline → Feature transforms: configures
fraud_transaction_detection/pipeline/feature.py - Feature pipeline → Data file I/O: saves features
fraud_transaction_detection/pipeline/feature.py - Feature transforms → Time strategies: checks strategies
fraud_transaction_detection/features/transformers.py - Pipeline initiator → Training pipeline: invokes
- Training pipeline → Data file I/O: loads features
fraud_transaction_detection/pipeline/training.py - Training pipeline → Model tuning: tunes models
fraud_transaction_detection/pipeline/training.py - Training pipeline → Model selection: selects model
fraud_transaction_detection/pipeline/training.py - Training pipeline → Training results: processes results
fraud_transaction_detection/pipeline/training.py - Training pipeline → Data file I/O: saves test data
fraud_transaction_detection/pipeline/training.py - Training results → Result interface: implements
fraud_transaction_detection/result_engine/training.py - Preprocessing pipeline → Pipeline interface: inherits
fraud_transaction_detection/pipeline/feature.py - Training pipeline → Pipeline interface: inherits
fraud_transaction_detection/pipeline/training.py - Deployment pipeline → Pipeline interface: inherits
fraud_transaction_detection/pipeline/deployment.py