ShipStack
← All projects
Machine Learning Intermediate

LoanLadder

Two-stage loan approval and amount estimation

Stage one decides approve or decline; stage two estimates how much to lend — with fairness checks across applicant groups.

SHIPSTACK · PROJECT LoanLadder Two-stage loan approval andamount estimation Intermediate Machine Learning 1 Application 2 Stage 1 3 Stage 2 4 Decision API 5 milestones

The scenario

LoanLadder chains a classifier and a regressor into a lending decision system. You will handle class imbalance, calibrate probabilities, check fairness metrics, and package the two models behind one API with input validation.

System architecture

01 Application Applicant data
02 Stage 1 Approve / decline
03 Stage 2 Loan amount
04 Decision API With reasons

High-level flow of LoanLadder, from input to output.

What you will learn

  • Chaining models into a decision system
  • Calibration and class imbalance
  • Fairness assessment

What you will build

  • Approval classifier
  • Amount regressor
  • Decision API

Build plan

  1. Data prep
  2. Classifier
  3. Regressor
  4. Fairness checks
  5. API

Tech stack

scikit-learnLightGBMFairlearnFastAPIDocker

More Machine Learning projects

See all →
SHIPSTACK · PROJECT ChurnShield Telecom customer churnprediction Beginner Machine Learning 1 Customer data 2 Features 3 Models 4 Retention app 5 milestones
Machine Learning Beginner

Telecom customer churn prediction

Predict which customers are likely to leave, explain why, and turn it into a retention list the business can act on.

pandas · scikit-learn · XGBoost Coming soon
SHIPSTACK · PROJECT StormCast Thunderstorm forecasting withMLflow tracking Beginner Machine Learning 1 Station data 2 Features 3 MLflow 4 Forecast API 5 milestones
Machine Learning Beginner

Thunderstorm forecasting with MLflow tracking

Forecast thunderstorm risk from weather-station data and track every experiment, parameter and model version with MLflow.

pandas · scikit-learn · MLflow Coming soon