Mustafa Cavus
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mustafa Cavus has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Przemyslaw Biecek
Faculty of Mathematics and Information Science, Warsaw University of Technology, Warsaw, Poland
7 shared publications Berna Yazici
4 shared publications- Berna Yazici
ESKİŞEHİR TEKNİK ÜNİVERSİTESİ, FEN FAKÜLTESİ, İSTATİSTİK BÖLÜMÜ
3 shared publications - Adrian Stando
Faculty of Mathematics and Information Science Warsaw University of Technology ul. Koszykowa 00-662 Warsaw Poland
2 shared publications - Ahmet Sezer
Bocconi School of Management
2 shared publications - Jakub Kuzilek
Humboldt-Universität zu Berlin, Berlin, Germany
2 shared publications
Latest Publications
- 11 May 2026
- Published by Elsevier in Future Generation Computer Systems
- Vol. 183
From XAI to MLOps: Explainable concept drift detection with the partial dependence profiles-based approach
- 11 May 2026
- Published by Elsevier in Future Generation Computer Systems
- Vol. 183
From XAI to MLOps: Explainable concept drift detection with the partial dependence profiles-based approach
- 7 May 2026
- Published by Springer Nature
- p. 3-18
An Experimental Study on the Rashomon Effect of Balancing Methods in Imbalanced Classification
- 7 May 2026
- Published by Springer Nature
- p. 3-18
An Experimental Study on the Rashomon Effect of Balancing Methods in Imbalanced Classification
- Open Access
- 13 April 2026
- Published by Springer Nature in Computational Economics
- p. 1-24
Explainable Bank Failure Prediction Models: Counterfactual Explanations to Reduce the Failure Risk
The accuracy and understandability of bank failure prediction models are crucial because inaccurate predictions may lead to undetected financial distress, resulting in economic losses. Likewise, models that are not interpretable hinder regulators and bank managers from taking timely and effective corrective actions. While interpretable models like logistic regression are favored for their explainability, complex models such as random forest, support vector machines, and deep learning offer higher predictive performance but lower explainability. These models, known as black boxes, make it difficult to derive actionable insights. To address this challenge, using counterfactual explanations is suggested. These explanations demonstrate how changes in input variables can alter the model’s output and suggest ways to mitigate bank failure risk. The key challenge lies in selecting the most effective method for generating useful counterfactuals, which should demonstrate validity , proximity , sparsity , plausibility , and diversity . The paper evaluates several model-agnostic counterfactual generation methods: What-If, Multi-Objective, and Nearest Instance Counterfactual Explanation, and also explores resampling methods like undersampling, oversampling, SMOTE, and the cost-sensitive approach to address data imbalance in bank failure prediction in the US. The results indicate that the Nearest Instance Counterfactual Explanation method yields higher-quality counterfactual explanations, mainly using the cost-sensitive approach. Overall, the Nearest Instance Counterfactual Explanation method outperforms others regarding validity , proximity , sparsity , and diversity metrics, with the cost-sensitive approach providing the most desirable counterfactual explanations. The combined use of Random Forest with cost-sensitive counterfactual generation further enhances both predictive accuracy and explanation quality, demonstrating a balanced and reliable framework for interpretable bank failure prediction. These findings highlight the variability in the performance of counterfactual generation methods across different balancing strategies and machine learning models, offering valuable strategies to enhance the utility of black-box bank failure prediction models.
- 13 April 2026
- Published by Springer Nature in Computational Economics
- p. 1-24
Open AccessExplainable Bank Failure Prediction Models: Counterfactual Explanations to Reduce the Failure Risk
The accuracy and understandability of bank failure prediction models are crucial because inaccurate predictions may lead to undetected financial distress, resulting in economic losses. Likewise, models that are not interpretable hinder regulators and bank managers from taking timely and effective corrective actions. While interpretable models like logistic regression are favored for their explainability, complex models such as random forest, support vector machines, and deep learning offer higher predictive performance but lower explainability. These models, known as black boxes, make it difficult to derive actionable insights. To address this challenge, using counterfactual explanations is suggested. These explanations demonstrate how changes in input variables can alter the model’s output and suggest ways to mitigate bank failure risk. The key challenge lies in selecting the most effective method for generating useful counterfactuals, which should demonstrate validity , proximity , sparsity , plausibility , and diversity . The paper evaluates several model-agnostic counterfactual generation methods: What-If, Multi-Objective, and Nearest Instance Counterfactual Explanation, and also explores resampling methods like undersampling, oversampling, SMOTE, and the cost-sensitive approach to address data imbalance in bank failure prediction in the US. The results indicate that the Nearest Instance Counterfactual Explanation method yields higher-quality counterfactual explanations, mainly using the cost-sensitive approach. Overall, the Nearest Instance Counterfactual Explanation method outperforms others regarding validity , proximity , sparsity , and diversity metrics, with the cost-sensitive approach providing the most desirable counterfactual explanations. The combined use of Random Forest with cost-sensitive counterfactual generation further enhances both predictive accuracy and explanation quality, demonstrating a balanced and reliable framework for interpretable bank failure prediction. These findings highlight the variability in the performance of counterfactual generation methods across different balancing strategies and machine learning models, offering valuable strategies to enhance the utility of black-box bank failure prediction models.
- Open Access
- 30 March 2026
- Published by The Open Journal in Journal of Open Source Software
- Vol. 11 (119) , 9481
datadriftR: An R package for data drift detection
- 30 March 2026
- Published by The Open Journal in Journal of Open Source Software
- Vol. 11 (119) , 9481
Open AccessdatadriftR: An R package for data drift detection
- Open Access
- 9 February 2026
- Published by Springer Nature in Information Systems Frontiers
Quantifying Model Uncertainty with AutoML and Rashomon Partial Dependence Profiles: Enabling Trustworthy and Human-centered XAI
Trustworthiness of AI systems is a core objective of Human-Centered Explainable AI, and relies, among other things, on explainability and understandability of the outcome. While automated machine learning tools automate model training, they often generate not only a single “best” model but also a set of near-equivalent alternatives, known as the Rashomon set. This set provides a unique opportunity for human-centered explainability: by exposing variability among similarly performing models, we can offer users richer and more informative explanations. In this paper, we introduce Rashomon partial dependence profiles , a model-agnostic technique that aggregates feature effect estimates across the Rashomon set. Unlike traditional explanations derived from a single model, Rashomon partial dependence profiles explicitly quantify uncertainty and visualize variability, further enabling user trust and understanding model behavior to make informed decisions. Additionally, under high-noise conditions, the Rashomon partial dependence profiles more accurately recover ground-truth feature relationships than a single-model partial dependence profile. Experiments on synthetic and real-world datasets demonstrate that Rashomon partial dependence profiles reduce average deviation from the ground truth by up to 38%, and their confidence intervals reliably capture true feature effects. These results highlight how leveraging the Rashomon set can enhance technical rigor while centering explanations on user trust and understanding aligned with Human-centered explainable AI principles.
- 9 February 2026
- Published by Springer Nature in Information Systems Frontiers
Open AccessQuantifying Model Uncertainty with AutoML and Rashomon Partial Dependence Profiles: Enabling Trustworthy and Human-centered XAI
Trustworthiness of AI systems is a core objective of Human-Centered Explainable AI, and relies, among other things, on explainability and understandability of the outcome. While automated machine learning tools automate model training, they often generate not only a single “best” model but also a set of near-equivalent alternatives, known as the Rashomon set. This set provides a unique opportunity for human-centered explainability: by exposing variability among similarly performing models, we can offer users richer and more informative explanations. In this paper, we introduce Rashomon partial dependence profiles , a model-agnostic technique that aggregates feature effect estimates across the Rashomon set. Unlike traditional explanations derived from a single model, Rashomon partial dependence profiles explicitly quantify uncertainty and visualize variability, further enabling user trust and understanding model behavior to make informed decisions. Additionally, under high-noise conditions, the Rashomon partial dependence profiles more accurately recover ground-truth feature relationships than a single-model partial dependence profile. Experiments on synthetic and real-world datasets demonstrate that Rashomon partial dependence profiles reduce average deviation from the ground truth by up to 38%, and their confidence intervals reliably capture true feature effects. These results highlight how leveraging the Rashomon set can enhance technical rigor while centering explanations on user trust and understanding aligned with Human-centered explainable AI principles.
- 12 January 2026
- Published by The R Foundation
ggskewboxplots: Skew Boxplot Geoms for 'ggplot2'
Provides 'ggplot2' extensions for creating skewed boxplots using several statistical methods (Kimber, 1990 ; Hubert and Vandervieren, 2008 ; Adil et al., 2015 ; Babura et al., 2017 ; Walker et al., 2018 ). The package implements custom statistical transformations and geometries to visualize data distributions with an emphasis on skewness.
- 12 January 2026
- Published by The R Foundation
ggskewboxplots: Skew Boxplot Geoms for 'ggplot2'
Provides 'ggplot2' extensions for creating skewed boxplots using several statistical methods (Kimber, 1990 ; Hubert and Vandervieren, 2008 ; Adil et al., 2015 ; Babura et al., 2017 ; Walker et al., 2018 ). The package implements custom statistical transformations and geometries to visualize data distributions with an emphasis on skewness.
- Open Access
ggskewboxplots: Enhanced Boxplots for Skewed Data in R
Traditional boxplots are widely used for summarizing and visualizing the distribution of numerical data, yet they exhibit significant limitations when applied to skewed or heavy-tailed distributions, often leading to misclassification of outliers through swamping -- flagging typical observations as outliers -- or masking -- failing to detect true outliers. This paper addresses these limitations by systematically evaluating several alternative boxplots specifically designed to accommodate distributional asymmetry. We introduce ggskewboxplots, an R package that integrates multiple robust and skewness-aware boxplot variants, providing a unified and user-friendly framework for exploratory data analysis. Using extensive Monte Carlo simulations under controlled skewness and kurtosis conditions, implemented via the mosaic approach based on the Skewed Exponential Power distribution, we assess the sensitivity and specificity of each method. Simulation results indicate that classical Tukey-style boxplots are highly prone to swamping and masking, whereas robust skewness-adjusted variants -- particularly those leveraging quartile-based skewness measures or medcouple-based adjustments -- achieve substantially better performance. These findings offer practical guidance for selecting reliable boxplot methods in applied settings and demonstrate how the ggskewboxplots package facilitates accessible, distribution-aware visualizations within the familiar ggplot2 workflow.
Open Accessggskewboxplots: Enhanced Boxplots for Skewed Data in R
Traditional boxplots are widely used for summarizing and visualizing the distribution of numerical data, yet they exhibit significant limitations when applied to skewed or heavy-tailed distributions, often leading to misclassification of outliers through swamping -- flagging typical observations as outliers -- or masking -- failing to detect true outliers. This paper addresses these limitations by systematically evaluating several alternative boxplots specifically designed to accommodate distributional asymmetry. We introduce ggskewboxplots, an R package that integrates multiple robust and skewness-aware boxplot variants, providing a unified and user-friendly framework for exploratory data analysis. Using extensive Monte Carlo simulations under controlled skewness and kurtosis conditions, implemented via the mosaic approach based on the Skewed Exponential Power distribution, we assess the sensitivity and specificity of each method. Simulation results indicate that classical Tukey-style boxplots are highly prone to swamping and masking, whereas robust skewness-adjusted variants -- particularly those leveraging quartile-based skewness measures or medcouple-based adjustments -- achieve substantially better performance. These findings offer practical guidance for selecting reliable boxplot methods in applied settings and demonstrate how the ggskewboxplots package facilitates accessible, distribution-aware visualizations within the familiar ggplot2 workflow.
- Open Access
- 19 October 2025
- Published by Springer Nature
- p. 3-26
Glocal Explanations of Expected Goal Models in Football
The expected goal models have gained popularity, but their interpretability is often limited, especially when trained using black-box methods. Explainable artificial intelligence tools have emerged to enhance model transparency and extract descriptive knowledge for a single observation or all observations. However, explaining black-box models for specific observations may be more useful in some domains. This paper introduces the glocal explanations (between local and global levels) of the expected goal models to enable performance analysis at the team and player levels by proposing aggregated versions of the SHAP values and partial dependence profiles. This allows knowledge to be extracted from the expected goal model for a player or team rather than just a single shot. In addition, we conducted real-data applications to illustrate the usefulness of aggregated SHAP and aggregated profiles. The paper concludes with remarks on the potential of these explanations for performance analysis in football analytics.
- 19 October 2025
- Published by Springer Nature
- p. 3-26
Open AccessGlocal Explanations of Expected Goal Models in Football
The expected goal models have gained popularity, but their interpretability is often limited, especially when trained using black-box methods. Explainable artificial intelligence tools have emerged to enhance model transparency and extract descriptive knowledge for a single observation or all observations. However, explaining black-box models for specific observations may be more useful in some domains. This paper introduces the glocal explanations (between local and global levels) of the expected goal models to enable performance analysis at the team and player levels by proposing aggregated versions of the SHAP values and partial dependence profiles. This allows knowledge to be extracted from the expected goal model for a player or team rather than just a single shot. In addition, we conducted real-data applications to illustrate the usefulness of aggregated SHAP and aggregated profiles. The paper concludes with remarks on the potential of these explanations for performance analysis in football analytics.
- 5 October 2025
- Published by Springer Nature
- p. 445-459
Beyond the Single-Best Model: Rashomon Partial Dependence Profile for Trustworthy Explanations in AutoML
- 5 October 2025
- Published by Springer Nature
- p. 445-459
Beyond the Single-Best Model: Rashomon Partial Dependence Profile for Trustworthy Explanations in AutoML