Nikita Kozodoi

Nikita Kozodoi
Humboldt-Universität zu Berlin | HU Berlin · Faculty of Economics and Business Administration

Doctor of Philosophy

About

9
Publications
3,015
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
87
Citations
Introduction
Machine Learning researcher & PhD candidate working on the frontier of research and business. My research focuses on ML applications in the field of credit risk analytics. My recent projects include: developing solutions to mitigate the impact of sampling bias on scoring models, performing profit-driven feature selection with multi-objective algorithms, reducing algorithmic discrimination and analyzing the profit-fairness trade-off. Check out my blog at https://kozodoi.me
Additional affiliations
April 2018 - present
Humboldt-Universität zu Berlin | Monedo
Position
  • Research Associate
Description
  • Doing research on ML applications in credit scoring in a joint collaboration between Monedo and Humboldt University of Berlin.
Education
October 2015 - November 2017
Humboldt-Universität zu Berlin
Field of study
  • Economics and Management Science
September 2014 - December 2016
September 2010 - July 2014

Publications

Publications (9)
Article
Full-text available
In credit scoring, feature selection aims at removing irrelevant data to improve the performance of the scorecard and its interpretability. Standard techniques treat feature selection as a single-objective task and rely on statistical criteria such as correlation. Recent studies suggest that using profit-based indicators may improve the quality of...
Conference Paper
Full-text available
Credit scoring models support loan approval decisions in the financial services industry. Lenders train these models on data from previously granted credit applications, where the borrowers’ repayment behavior has been observed. This approach creates sample bias. The scoring model is trained on accepted cases only. Applying the model to screen appl...
Preprint
Full-text available
The rise of algorithmic decision-making has spawned much research on fair machine learning (ML). Financial institutions use ML for building risk scorecards that support a range of credit-related decisions. Yet, the literature on fair ML in credit scoring is scarce. The paper makes two contributions. First, we provide a systematic overview of algori...
Thesis
Der Aufstieg des maschinellen Lernens (ML) und die rasante Digitalisierung der Wirtschaft haben die Entscheidungsprozesse in der Finanzbranche erheblich verändert. Finanzinstitute setzen zunehmend auf ML, um die Entscheidungsfindung zu unterstützen. Kreditscoring ist eine der wichtigsten ML-Anwendungen im Finanzbereich. Die Aufgabe von Kreditscorin...
Article
Full-text available
The rise of algorithmic decision-making has spawned much research on fair machine learning (ML). Financial institutions use ML for building risk scorecards that support a range of credit-related decisions. Yet, the literature on fair ML in credit scoring is scarce. The paper makes three contributions. First, we revisit statistical fairness criteria...
Chapter
Full-text available
Credit scoring refers to the use of statistical models to support loan approval decisions. An ever-increasing availability of data on potential borrowers emphasizes the importance of feature selection for scoring models. Traditionally, feature selection has been viewed as a single-objective task. Recent research demonstrates the effectiveness of mu...
Experiment Findings
In this experiment, we simulate the acceptance loop using synthetic data and compare three scoring models: 1) Accepts-based model that ignores rejected cases; 2) Model that relies on the augmented data from using reject inference with shallow self-learning; 3) Oracle model that depicts the best linear separator in the feature space. Our experiment...
Preprint
Full-text available
Credit scoring models support loan approval decisions in the financial services industry. Lenders train these models on data from previously granted credit applications, where the borrowers' repayment behavior has been observed. This approach creates sample bias. The scoring model (i.e., classifier) is trained on accepted cases only. Applying the r...
Conference Paper
Full-text available
In credit scoring, feature selection aims at removing irrelevant data to improve the performance of the scorecard and its interpretability. Standard feature selection techniques are based on statistical criteria such as correlation. Recent studies suggest that using profit-based indicators for model evaluation may improve the quality of scoring mod...

Network

Cited By

Projects

Projects (2)
Project
Credit scoring models are trained on data from previously accepted applicants, which creates sampling bias. In this project, we develop new reject inference techniques that overcome bias through assigning labels to rejected cases.
Project
Standard feature selection techniques rely on statistical criteria to select features. In this project, we investigate the usage of profit measures to perform profit-driven feature selection for credit scoring.