B1674
Title: Statistical learning from data
Authors: Seitebaleng Makgai - University of Pretoria (South Africa) [presenting]
JT Ferreira - University of Pretoria (South Africa)
Andriette Bekker - University of Pretoria (South Africa)
Abstract: As an important indicator of social well-being, it is essential to have a comprehensive understanding of the distribution of population income for policy-makers to reduce harmful social and economic effects. The lognormal distribution is often considered the model of choice for the modelling of income data. This model is utilized in a data-enriching way; by using firstly a regression curve in a condition-unconditional approach. The regression coefficient inherits the dependence structure and behaves as an inequality parameter. By doing so, income heterogeneity is interpreted, and findings emphasize that this approach can be successfully used in practice for health- and other socio-economic data scenarios, amongst others. Furthermore, this presentation presents a novel extension where the data enrichment enters the conditional mode parametrized lognormal model via the mode curve for fitting income data. A detailed simulation study illustrates the performance of the models and the value added by this different perspective and implementation. A data set application is included to illustrate this approach.
