TY - THES TI - Reversing the Coarse : Statistical Methods for Recovering Structure from Aggregated, Sparse, and Discretized Data AU - Mühlbauer, Michael PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-117096 UR - https://fis.uni-bamberg.de/handle/uniba/117096 LA - eng AB - Empirical research often works with data coarser than the question demands: a survey too sparse for district-level estimates, geocoordinates collapsed into regional totals, a continuous preference recorded on a five-point scale. This cumulative dissertation treats these as one problem — coarsened data — and recovers fine-grained structure by combining the coarse observation with auxiliary information. Chapter 1 estimates district-level person-kilometers traveled from the German household travel survey Mobilität in Deutschland 2017 with a log-transformed Fay-Herriot model across 1,590 domains. Covariates from official statistics and the INKAR database are selected in a three-phase LASSO procedure, reducing coefficients of variation by 16–44% against direct estimation. Chapter 2 extends kernel density estimation from aggregated data by integrating auxiliary spatial information through a correlation-based weighting scheme. The estimator is assessed in model-based simulations and in an evaluation study on population densities from the 2022 German Census in Bavaria, and then applied to brown hare hunting bag data in Lower Saxony. Chapter 3 applies the mixed-effects frequency-adjusted borders ordinal forest to lexical preference ratings in Maltese English and contributes partial dependence routines in R. It recovers curved, non-monotonic age trajectories that the cumulative link mixed model does not. KW - Coarsened data KW - Small area estimation KW - Fay-Herriot model KW - Kernel density estimation KW - Ordinal data KW - Random forests ER - TY - THES TI - Generalized Tree-Based Machine Learning Methods with Applications to Small Area Estimation AU - Frink, Nicolas PY - 2025 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-107660 UR - https://fis.uni-bamberg.de/handle/uniba/107660 LA - eng AB - Chapter 1 - Identifying and addressing poverty is challenging in administrative units with limited information on income distribution and well-being. To overcome this obstacle, small area estimation methods have been developed to provide reliable and efficient estimators at disaggregated levels, enabling informed decision-making by policymakers despite the data scarcity. We propose a robust and flexible approach for estimating poverty indicators based on binary response variables within the small area estimation context: the generalized mixed effects random forest. Our method employs machine learning techniques to identify predictive, non-linear relationships from data, while also modeling hierarchical structures. Mean squared error estimation is explored using a parametric bootstrap. From an applied perspective, we examine the impact of information loss due to converting continuous variables into binary variables on the performance of small area estimation methods. We evaluate the proposed point and uncertainty estimates in both model- and design-based simulations. Finally, we apply our method to a case study revealing spatial patterns of poverty in the Mexican state of Tlaxcala. KW - Chapter 1 - Data integration, Generalized mixed models, MSE estimation, Parametric bootstrap KW - Chapter 2 - Bootstrap, Generalized linear mixed models, Overdispersed count data, Random forest, Small area estimation KW - Chapter 3 - Generalized linear mixed models, Machine learning, Small area estimation, Survey statistics ER - TY - THES TI - A Statistical Learning Framework for Multi-Target Prediction of Image Processing Parameters : An Empirical Application for Head-up-Display Calibration in Automotive Production AU - Remlein, Gloria Maria PY - 2025 PB - Eigenverlag Gloria Maria Remlein CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/107569 LA - eng ER - TY - THES TI - Recent Advances in Small Area Estimation of Economic and Poverty Indicators using Traditional and Alternative Data AU - Lee, Yeonjoo PY - 2024 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-103063 UR - https://fis.uni-bamberg.de/handle/uniba/103063 LA - eng AB - Chapter 1 - Variable selection using conditional AIC for linear mixed models with data-driven transformations When data analysts use linear mixed models, they usually encounter two practical problems: a) the true model is unknown and b) the Gaussian assumptions of the errors do not hold. While these problems commonly appear together, researchers tend to treat them individually by a) finding an optimal model based on the conditional Akaike information criterion (cAIC) and b) applying transformations on the dependent variable. However, the optimal model depends on the transformation and vice versa. In this paper, we aim to solve both problems simultaneously. In particular, we propose an adjusted cAIC by using the Jacobian of the particular transformation such that various model candidates with differently transformed data can be compared. From a computational perspective, we propose a step-wise selection approach based on the introduced adjusted cAIC. Model-based simulations are used to compare the proposed selection approach to alternative approaches. Finally, the introduced approach is applied to Mexican data to estimate poverty and inequality indicators for 81 municipalities. KW - Box-Cox transformation KW - Empirical best predictor KW - Indicators KW - Small area estimation KW - inflation KW - regional inequality KW - regional difference KW - price statistics KW - household consumption KW - head count ratio KW - Mozambique KW - satellite imagery KW - remote sensing data KW - subarea level model ER -