Relationship between monthly income and chosen explanatory variables: example of regression analysis with emphasis on its possible goals

The diploma thesis deals with regression analysis with emphases on its possible goals. For practical part of the thesis was chosen dataset on which it was shown how regression analysis should be treated according to selected goal. If the goal of regression analysis is to describe analysed dataset, no statistical hypothesis should be run. On the other hand, no theoretical assumptions are needed to check for a dataset description, so description can be always done. To generalizing relationships between variables to the population or data-generated process, it is necessary to meet assumptions about error term. If these assumptions are not satisfied, no hypothesis testing or constructing confidence intervals should be made. For causal goal of regression analysis even more assumptions need to be satisfied, which is very hard to meet. If these assumptions are not satisfied, regression parameters should not be interpreted causally. Prediction of regression analysis do not require any assumptions, its aim is to find the best regression function or machine learning algorithm, which will best predict response variable according to new values of explanatory variables. All regression analyses in the thesis use variables as monthly income, age, total working years and education. Response variable is always monthly income. Difference between paired and partial relationships are also shown in this thesis. To achieve the goals of the regression analysis, the statistical program Rstudio is used, which enables to show all the graphs and calculate/estimate regression parameters in this thesis.