Ocena zdolności kredytowej (scoring)

Credit scoring and underwriting · id creditscoring · aliases: credit scoring, credit risk modelling, underwriting, PD model, scoring kredytowy, ryzyko kredytowe

Modele prawdopodobieństwa niewypłacalności (PD) i decyzje kredytowe: od kart scoringowych po gradient boosting, z wymogami wyjaśnialności i niedyskryminacji.

Probability-of-default (PD) models and credit decisions: from scorecards to gradient boosting, under explainability and non-discrimination requirements.

In one sentence

Modele, które na podstawie twojej historii finansowej szacują, czy spłacisz pożyczkę, i pomagają bankowi zdecydować o kredycie.

Models that use your financial history to estimate whether you will repay a loan and help the bank decide.

Problem it solves

Bank musi szybko zdecydować, komu pożyczyć i na jakich warunkach, nie znając osobiście tysięcy wnioskodawców.

A bank must decide quickly whom to lend to and on what terms, without knowing thousands of applicants personally.

What it does

Jak doradca kredytowy, który pamięta losy setek tysięcy wcześniejszych pożyczek i na tej podstawie ocenia nową.

Like a loan officer who remembers how hundreds of thousands of earlier loans turned out and judges the new one by that.

Everyday example

Wypełniasz w sklepie internetowym wniosek o raty na telefon i po kilku minutach widzisz decyzję oraz proponowaną kwotę.

You fill in an instalment application for a phone in an online shop and within minutes see the decision and offered amount.

This concerns you if you…

starasz się o kredyt, raty lub kartę kredytową
dostałeś odmowę i nie wiesz dlaczego
budujesz modele ryzyka w banku lub fintechu
apply for a loan, instalments or a credit card
were turned down and don't know why
build risk models at a bank or fintech

Subtopics

Karty scoringowe i modele PD (Credit scorecards and PD models): Logistic regression on binned features encoded by weight of evidence (WoE), rescaled to points; the reference pattern for credit scoring.
Modele straty przy niewypłacalności (LGD) i oczekiwanej straty kredytowej (Loss given default (LGD) and expected credit loss models): LGD and EAD models that combine with PD into expected loss (EL = PD×LGD×EAD) for regulatory capital and IFRS 9 / CECL reserves.
Ocena kredytowa z historii rachunku (dane alternatywne) (Cash-flow underwriting with alternative data): Credit scoring on features from account transaction data (inflows, balances, obligations) obtained with consent, e.g. via open banking.
Wnioskowanie o odrzuconych wnioskach (reject inference) (Reject inference in credit scoring): Methods correcting sample selection bias in scoring by assigning inferred outcomes to rejected applications before model training.
Wyjaśnianie decyzji kredytowych i równe traktowanie (Explaining credit decisions and fair lending): Adverse-action reason codes, feature attribution (SHAP), counterfactual explanations and discrimination testing for regulated credit models.

Sources

Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research · Stefan Lessmann, Bart Baesens, Hsin-Vonn Seow, Lyn C. Thomas
EU AI Act – Annex III: High-risk AI systems (point 5(b) creditworthiness, 5(c) life and health insurance) · European Parliament and Council of the European Union