Robust instance-dependent cost-sensitive classification

Article Computer Science, Artificial Intelligence

Cost-sensitive ensemble learning: a unifying framework

George Petrides et al.

Summary: The study introduces a unified framework for cost-sensitive ensemble methods, categorizing and comparing them, including extensions and generalizations for methods like AdaBoost, Bagging, and Random Forest.

DATA MINING AND KNOWLEDGE DISCOVERY (2022)

添加到收藏夹

Article Management

Cost-sensitive learning for profit-driven credit scoring

George Petrides et al.

Summary: This study evaluates the effectiveness of cost-sensitive learning methods in improving the profitability of credit scorecards. The results show that these methods can increase profitability, especially for channels with higher default rates.

JOURNAL OF THE OPERATIONAL RESEARCH SOCIETY (2022)

添加到收藏夹

Article Management

Instance-dependent cost-sensitive learning for detecting transfer fraud

Sebastiaan Hoppner et al.

Summary: Credit card transaction fraud is a global issue, and financial institutions are increasingly relying on data-driven methods to develop fraud detection systems for detecting and preventing fraudulent transactions. The article introduces two novel classifiers that minimize the instance-dependent cost measure when learning a classification model, highlighting the potential to reduce fraud losses.

EUROPEAN JOURNAL OF OPERATIONAL RESEARCH (2022)

添加到收藏夹

Article Computer Science, Information Systems

Predict-then-optimize or predict-and-optimize? An empirical evaluation of cost-sensitive learning strategies

Toon Vanderschueren et al.

Summary: Predictive models are increasingly used to optimize decision-making and minimize costs. This work compared the predict-then-optimize approach with the predict-and-optimize approach in cost-sensitive classification. The key finding was that the decision-making strategy was generally more effective than training with a task-specific loss or their combination.

INFORMATION SCIENCES (2022)

添加到收藏夹

Article Statistics & Probability

Robust logistic zero-sum regression for microbiome compositional data

G. S. Monti et al.

Summary: The Robust Logistic Zero-Sum Regression (RobLZS) estimator introduced in this study is designed for two-class problems with high-dimensional compositional covariates, utilizing a log-contrast model for feature selection and robustness. Comparative performance evaluations using Monte Carlo simulations and microbiome data applications demonstrate the stability and effectiveness of the RobLZS estimator in handling outliers and conducting feature selection. The availability of the Robust Logistic Zero-Sum Regression as an R package offers a practical tool for researchers in the field.

ADVANCES IN DATA ANALYSIS AND CLASSIFICATION (2022)

添加到收藏夹

Article Computer Science, Information Systems

Targeting customers for profit: An ensemble learning framework to support marketing decision-making

Stefan Lessmann et al.

Summary: The study highlights the importance of targeting the right customer groups in marketing campaigns and proposes a profit-conscious ensemble selection modeling framework that integrates statistical learning principles with business objectives, contributing to the field of profit analytics.

INFORMATION SCIENCES (2021)

添加到收藏夹

Article Statistics & Probability

Robust regression with compositional covariates including cellwise outliers

Nikola Stefelova et al.

Summary: The study presents a robust procedure for estimating a linear regression model with compositional and real-valued explanatory variables, designed to handle outliers and produce results aligned with established scientific knowledge. By filtering and imputing cellwise outliers before performing rowwise robust compositional regression, the proposed procedure outperforms traditional and other robust regression methods.

ADVANCES IN DATA ANALYSIS AND CLASSIFICATION (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence