learning program

MACHINE Learning for Scoring Models

A structured set of courses covering the methods, tools, and real-world applications behind credit scoring and risk classification using machine learning. Each course targets a specific stage of that workflow.

all courses

AVAILABLE tracks

ML Scoring Models: From Raw Data to Credit Decisions Machine Learning Intermediate

ML Scoring Models: From Raw Data to Credit Decisions

A technical course covering the full pipeline for building machine learning scoring models — from feature engineering to model validation and regulatory compliance.

9 min
5 weeks
14 seats left
Credit Scoring with ML: A Structured Introduction Credit Risk Beginner

Credit Scoring with ML: A Structured Introduction

For analysts stepping into data science — learn to build interpretable scoring models without prior ML experience, using Python and real lending datasets.

7 min
5 weeks
22 seats left
Advanced Risk Scoring: Gradient Boosting, Fairness, and Production MLOps & Risk Professional

Advanced Risk Scoring: Gradient Boosting, Fairness, and Production

A practitioner-level program on deploying gradient boosting scoring models in regulated financial environments — covering fairness auditing, drift detection, and MLOps basics.

11 min
5 weeks
8 seats left
Behavioral Scoring Models: Predicting Risk Over Time Behavioral Analytics Intermediate

Behavioral Scoring Models: Predicting Risk Over Time

Learn to build behavioral scoring models using account transaction history, payment patterns, and time-series features — distinct from application scoring in both method and data.

10 min
5 weeks
11 seats left
Alternative Data in ML Scoring: Methods and Limitations Alternative Data Intermediate / Advanced

Alternative Data in ML Scoring: Methods and Limitations

A focused course on integrating non-traditional data sources — telco, open banking, rental history — into credit scoring models, with honest treatment of data quality and regulatory risk.

10 min
5 weeks
9 seats left

how courses are built

what the PROGRAM covers

Each course in the program targets a specific phase of scoring model development — from raw data preparation through to model validation and deployment. The curriculum draws on real lending datasets and risk classification scenarios rather than synthetic examples.

Core curriculum checklist

4 of 6 topics covered across all tracks
Feature engineering for credit variables — handling missingness, encoding categorical fields, constructing ratio features
Logistic regression and scorecard construction — weight of evidence, information value, binning strategies
Tree-based methods — gradient boosting with XGBoost and LightGBM applied to binary default prediction
Model evaluation — Gini coefficient, KS statistic, ROC-AUC, calibration curves for probability outputs
Reject inference — handling sample bias from historical approval decisions in training data
Model monitoring and drift detection — PSI, CSI, and scheduled recalibration workflows

Technical prerequisites covered

3 of 4 topics addressed
Python fundamentals — pandas, NumPy, and scikit-learn pipelines used throughout exercises
SQL basics — pulling structured loan data from relational tables for feature construction
Statistical foundations — probability distributions, hypothesis testing, correlation analysis
MLOps basics — model versioning, experiment tracking with MLflow, deployment patterns
Machine learning scoring model workflow in practice
applied scoring

about the program

scoring MODELS built from actual data

The program started in 2020 with the observation that most ML courses teach classification in abstract terms — accuracy on toy datasets, not Gini coefficients on real loan portfolios. Courses here use credit bureau files, loan origination records, and repayment histories as the working material throughout. Participants work in Python environments that mirror production risk team setups, not cleaned academic benchmarks.

5
course tracks across the scoring workflow
6
core ML topics addressed across all tracks
4
distinct experience levels served
online
fully remote, accessible nationally
Your data preferences

We use data to keep things running and to improve your quiz experience: