Machine Learning on Pi.Kappa_

Entity Matching with Similarity Encoding: A Supervised Learning Recommendation Framework for Linking (Big) Data

Mon, 01 Jan 0001 00:00:00 +0000

Karapanagiotis, P., & Liebald, M. (2023). Entity Matching with Similarity Encoding: A Supervised Learning Recommendation Framework for Linking (Big) Data. SAFE Working Paper Series. No. 398. https://doi.org/10.2139/ssrn.4541376.

In this study, we introduce a novel entity matching (EM) framework. It com-bines state-of-the-art EM approaches based on Artiﬁcial Neural Networks (ANN) with a new similarity encoding derived from matching techniques that are preva-lent in ﬁnance and economics. Our framework is on-par or outperforms alternative end-to-end frameworks in standard benchmark cases. Because similarity encod-ing is constructed using (edit) distances instead of semantic similarities, it avoids out-of-vocabulary problems when matching dirty data. We highlight this property by applying an EM application to dirty ﬁnancial ﬁrm-level data extracted from historical archives.

Face Reading Technology: Improving Preference Prediction from Self-Reports Using Micro Expressions

Mon, 01 Jan 0001 00:00:00 +0000

Karapanagiotis, P., Krause, F., & Krick, J. (2026). Face Reading Technology: Improving Preference Prediction from Self-Reports Using Micro Expressions. In Journal of Business Research (Vol. 208). https://doi.org/10.1016/j.jbusres.2026.116050

Traditional market research primarily relies on self-reports (SR) to assess consumer preferences, yet these methods are prone to biases and limited in capturing subconscious emotional responses. Psychophysiological and neurophysiological methods offer objective alternatives, but their high cost and intrusiveness limit practical use. This study examines automated Facial Expression Analysis (FEA), focusing on micro-expression (ME) emotion data, as a scalable, non-intrusive approach to improve the accuracy of predicting consumer choices. In a controlled experiment exposing participants to both video and poster ads, we compare the predictive power of ME and SR emotion data using machine learning and artificial neural network models. Results demonstrate that ME data significantly enhance both multinomial and binomial choice prediction accuracy compared to SR data, particularly for dynamic video ads where ME patterns capture real-time emotional fluctuations more effectively. Beyond its methodological contributions, our study underscores practical implications and discusses the ethical considerations related to consumer privacy.

Coding and Data Visualization with Generative AI

Mon, 01 Jan 0001 00:00:00 +0000

This course introduces the fundamental computer science concepts and techniques relevant to managers. The course aims to equip students with the essential knowledge and skills needed to apply and understand computational methods to economic analysis, modeling, and data analysis. Previous experience with data science can be helpful, but it is not required. The course begins with an overview of programming concepts and common activities that are relevant to data science. Then, it delves into advanced topics in programming, visualization, and data management.