How sonalpriyanshi529/loanapprovalprediction works
This repository trains and serves a binary loan-approval classifier. The main user workflow is applicant details in the Streamlit app → feature preparation → model prediction and approval probabilities. A separate training workflow loads loan data, preprocesses and engineers features, trains and evaluates several models, then saves artifacts consumed by the app. The README also documents notebook-based EDA and model comparisons; notebook internals are not sampled, so only its documented role is shown.
Main entry points: train_model.py Training orchestrator · app.py Streamlit interface · Model Model artifacts · loan_data.csv Loan dataset
Components 13
Analysis
- EDA and comparison
Notebook/Loan_Approval_Prediction.ipynb
Training
- Loan dataset
Dataset/loan_data.csv - Training orchestrator
train_model.py - Preprocessing and features
train_model.py - Model training
train_model.py - Model evaluation
train_model.py - Model artifacts
Model
Prediction app
- Streamlit interface
Streamlit_App/app.py - Artifact loading
Streamlit_App/app.py - Input feature building
Streamlit_App/app.py - Approval inference
Streamlit_App/app.py - Decision and probabilities
Streamlit_App/app.py
Other components
- Applicant
Connections 17
- EDA and comparison → Loan dataset: analyzes
- Training orchestrator → Loan dataset: loads
- Training orchestrator → Preprocessing and features: calls
- Preprocessing and features → Model training: provides features
- Training orchestrator → Model training: calls
- Training orchestrator → Model evaluation: calls
- Model training → Model evaluation: evaluated by
- Training orchestrator → Model artifacts: saves
- Model evaluation → Model artifacts: metrics saved
- Applicant → Streamlit interface: enters details
- Streamlit interface → Artifact loading: loads models
- Artifact loading → Model artifacts: reads
- Streamlit interface → Input feature building: submits inputs
- Input feature building → Approval inference: provides features
- Streamlit interface → Approval inference: requests prediction
- Approval inference → Decision and probabilities: returns decision
- Streamlit interface → Decision and probabilities: displays