Logo Logo
Hilfe
Kontakt
Switch language to English
Differential human factors in user data
Differential human factors in user data
This thesis investigates how differential human factors, such as demography and personality, are related to actual individual behavior. Within this broad context, this work addresses the prevailing lack of real behavior in the scientific field of psychology and differential-/social psychology in particular. Furthermore, this work provides an introduction to the practice of data-logging as a promising alternative to self-reports for the collection of behavioral data. Additionally, we introduce new data-analytical concepts from the field of machine learning in order to appropriately handle large and noisy datasets, such as technical logs. To illustrate these concepts we provide three empirical studies, using behavioral logging procedures. In the first study we report on data obtained in a virtual automotive driving simulation. Using these data, we demonstrate how individual driving patterns can be used to predict driver gender with high accuracy from basic automotive driving logs. Additionally, we provide information about the most important variables associated with male and female driving styles. Two additional studies utilize a specially designed Android application, to automatically collect behavioral user data in a privacy protecting manner from participants private smartphones. The second study describes how most stable mobile application usage on smartphones can be predicted from individual personality and demography scores and highlights implications for personality sensitive recommender systems. The third study demonstrates how individual personality can potentially be predicted, using a wide range of user interactions, with a machine learning approach. Finally, we discuss the reported results within the context of previous research and highlight possible implications of technological advancements for psychological science.
Data Logging, Machine Learning, Personality, Behavior, Smartphone Usage
Stachl, Clemens
2017
Englisch
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Stachl, Clemens (2017): Differential human factors in user data. Dissertation, LMU München: Fakultät für Psychologie und Pädagogik
[thumbnail of Stachl_Clemens.pdf]
Vorschau
PDF
Stachl_Clemens.pdf

1MB

Abstract

This thesis investigates how differential human factors, such as demography and personality, are related to actual individual behavior. Within this broad context, this work addresses the prevailing lack of real behavior in the scientific field of psychology and differential-/social psychology in particular. Furthermore, this work provides an introduction to the practice of data-logging as a promising alternative to self-reports for the collection of behavioral data. Additionally, we introduce new data-analytical concepts from the field of machine learning in order to appropriately handle large and noisy datasets, such as technical logs. To illustrate these concepts we provide three empirical studies, using behavioral logging procedures. In the first study we report on data obtained in a virtual automotive driving simulation. Using these data, we demonstrate how individual driving patterns can be used to predict driver gender with high accuracy from basic automotive driving logs. Additionally, we provide information about the most important variables associated with male and female driving styles. Two additional studies utilize a specially designed Android application, to automatically collect behavioral user data in a privacy protecting manner from participants private smartphones. The second study describes how most stable mobile application usage on smartphones can be predicted from individual personality and demography scores and highlights implications for personality sensitive recommender systems. The third study demonstrates how individual personality can potentially be predicted, using a wide range of user interactions, with a machine learning approach. Finally, we discuss the reported results within the context of previous research and highlight possible implications of technological advancements for psychological science.