Fabian Obster

Fabian Obster

Auto-generated profile, based on publicly available information

Fabian Obster

Department of Business Administration, University of the Bundeswehr Munich, Neubiberg, Germany

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 April 2022 in Policy & Internet
Verified
Journal Article
Open access
NTNeil Thurman
Neil Thurman
ANAsmik Nalmpatian
Asmik Nalmpatian
FOFabian Obster
Fabian Obster
In 2020, the French Parliament passed an amendment that put the country at the forefront of attempts by democratic states to restrict young people's access to legal online pornography. This study examines the necessity for and potential efficacy of the amendment, Article 23, through a comparative analysis of emerging legislative and regulatory approaches in France, the UK, Canada, Utah, and Germany, and through a survey of French 15‐, 16‐, and 17‐year‐olds. Among other things, our survey shows that 41% of 15‐, 16‐, and 17‐year‐olds in France visit dedicated pornographic sites, on average monthly and often much more frequently. However, the range of media platforms via which French adolescents are exposed to pornography, their knowledge about technologies that could circumvent age verification, and the power, scope, and implementation of Article 23 may limit the legislation's efficacy. Our findings suggest the mechanisms that may limit its efficacy include media displacement, socio‐technical circumvention, and the Article's relatively broad and imprecise nature. This study has implications for legislators and regulators in democratic countries beyond France as they too grapple with the challenges of regulating online pornography. Furthermore, it extends the often contradictory and/or limited evidence that exists about adolescents' consumption of pornography.

5 January 2024 in Communications Earth & Environment
Verified
Journal Article
Open access
PDF file
FOFabian Obster
Fabian Obster
HBHeidi Bohle
Heidi Bohle
PPPaul M. Pechan
Paul M. Pechan
Climate change has significant implications for economically important crops, yet understanding its specific impact on farm financial wellbeing remains a challenging task. In this study we present self-reported perceptions of fruit farmers about their financial well-being when confronted with different climate change factors. We employed a combination of supervised machine learning and statistical modelling methods to analyze the data. The data collection was conducted through face-to-face interviews with 801 randomly selected cherry and peach farmers in Tunisia and Chile. Specific climate change factors, namely increases in temperature and reductions in precipitation, can have a regionally discernible effect on the self-perceived financial wellbeing of fruit farmers. This effect is less pronounced in Tunisia than in Chile. However, climate change is of lessor importance in predicting farm financial wellbeing, particularly for farms already doing well financially. Social assets, which include reliance on and trust in information sources, community and science, play an important role in increasing the probability of fruit farm financial wellbeing in both Tunisia and Chile. However, the most influential predictive factors differ between the two countries. In Chile, the location of the farm is the primary determinant of financial wellbeing, while in Tunisia it was the presence of social assets.

7 August 2023 in Scientific Reports
Verified
Journal Article
Open access
PDF file
FOFabian Obster
Fabian Obster
CHChristian Heumann
Christian Heumann
HBHeidi Bohle
Heidi Bohle
PPPaul Pechan
Paul Pechan
We describe how interpretable boosting algorithms based on ridge-regularized generalized linear models can be used to analyze high-dimensional environmental data. We illustrate this by using environmental, social, human and biophysical data to predict the financial vulnerability of farmers in Chile and Tunisia against climate hazards. We show how group structures can be considered and how interactions can be found in high-dimensional datasets using a novel 2-step boosting approach. The advantages and efficacy of the proposed method are shown and discussed. Results indicate that the presence of interaction effects only improves predictive power when included in two-step boosting. The most important variable in predicting all types of vulnerabilities are natural assets. Other important variables are the type of irrigation, economic assets and the presence of crop damage of near farms.

23 December 2024 in IEEE Access
Verified
Journal Article
Open access
FOFabian Obster
Fabian Obster
MCMonica Ciolacu
Monica Ciolacu
Andreas HumpeAndreas Humpe
Andreas Humpe
This paper investigates the empirical relationship between predictive performance, often called predictive power, and interpretability of various Machine Learning algorithms, focusing on bicycle traffic data from four cities. As Machine Learning algorithms become increasingly embedded in decision-making processes, particularly for traffic management and other high-level commitment applications, concerns regarding the transparency and trustworthiness of complex ’black-box’ models have grown. Theoretical assertions often propose a trade-off between model complexity (predictive performance) and transparency (interpretability); however, empirical evidence supporting this claim is limited and inconsistent. To address this gap, we introduce a novel interpretability scoring system - a Machine Learning Interpretability Rank-based scale - that combines objective measures such as the number of model parameters with subjective interpretability rankings across different model types. This comprehensive methodology includes stratified sampling, model tuning, and a two-step ranking system to operationalize this trade-off. Results reveal a significant negative correlation between interpretability and predictive performance for intrinsically interpretable models, reinforcing the notion of a trade-off. However, this relationship does not hold for black-box models, suggesting that for these algorithms, predictive performance can be prioritized over interpretability. This study contributes to the ongoing discourse on explainable AI, providing practical insights and tools to help researchers and practitioners achieve a balance between model complexity and transparency. We recommend to prioritise more interpretable models when predictive performance is comparable. Our scale provides a transparent and efficient framework for implementing this heuristic and improving parameter optimization. Further research should extend this analysis to unstructured data, explore different interpretability methods, and develop new metrics for evaluating the trade-off across diverse contexts.

14 November 2024 in The American Statistician
Verified
Journal Article
Open access
FOFabian Obster
Fabian Obster
CHChristian Heumann
Christian Heumann

3 August 2021 in bioRxiv
Verified
Preprint
Open access
PDF file
ABAnna M. Biller
Anna M. Biller
CMCarmen Molenda
Carmen Molenda
FOFabian Obster
Fabian Obster
GZGiulia Zerbini
Giulia Zerbini
CFChristian Förtsch
Christian Förtsch
EWEva Winnebeck
Eva Winnebeck
The mismatch between teenagers’ late sleep phase and early school start time results in acute and chronic sleep reductions. This is not only harmful for students’ learning in the short-term but may impact on students’ career prospects and widen social inequalities. Delaying school start times has been shown to improve sleep but whether this translates to better achievement is unresolved. The current evidence is limited due to a plethora of outcome measures and the many factors influencing sleep and grade/score trajectories. Here, we studied whether 0.5-1.5 years of exposure to a flexible school start system, with the daily choice of an 8AM or 8:50AM-start (intervention), allowed secondary school students (n=63-157, 14-19 years) to improve their quarterly school grades in a 4-year longitudinal pre-post design. We investigated whether sleep, changes in sleep or frequency of later starts predicted grade improvements in the flexible system. Our mixed model regressions with 5,111-16,724 official grades as outcomes did not indicate meaningful grade improvements in the flexible system per se or with previously observed sleep variables (nor their changes) – the covariates academic quarter, discipline and grade level had a greater, more systematic effect in our sample. Importantly, this finding does not preclude improvements in learning and cognition in our sample. However, at the ‘dose’ received here, intermittent sleep benefits did not obviously translate into detectable grade changes, which is in line with several other studies and highlights that grades are suboptimal to evaluate timetabling interventions despite their importance for future success. Significance statement: Early school start times worldwide clash with teenagers’ delayed sleep-wake rhythms. This mismatch results in sleep restrictions below healthy amounts, which compromises health and performance and further aggravates social disparities. Since adequate sleep is important for learning and concentration, there is the strong expectation that counteracting sleep deprivation with delayed school starts results in better academic achievement. We add important high-resolution, longitudinal data to this unresolved scientific debate. When controlling for confounders, our results do not support that improved sleep leads to grade changes within 1.5 years in a flexible start system. While grades are suboptimal to measure later school start effects on performance, they nevertheless open doors to higher education worldwide and thus determine future trajectories.