Modern approaches to imputing missing values in categorical data

The bachelor’s thesis focuses on modern approaches to imputing missing values in categorical data. The aim of the thesis is to compare the efficiency and accuracy of the MICE and EM algorithms in handling missing categorical data across various scenarios. These advanced methods are also compared with simple imputation methods to assess their added value in comparison to less demanding solutions. In addition to describing the basic terminology, types of missing data mechanisms, and approaches to solve them, the thesis also focuses on the implementation of the MICE and EM algorithms in the R environment. These methods are applied to real datasets with simulated missing values of different types. In the experimental section, the results of missing data imputation are evaluated using appropriate metrics, taking into account the type of missingness and the structure of the categorical data. The conclusions of the thesis provide recommendations for selecting the optimal imputation method based on the characteristics of the dataset and the type of missing values. Moreover the thesis can assist researchers and analysts in working with incomplete categorical data.