Investigation of turkish labor force characteristics: logistic regression and decision tree application

The purpose of this thesis is threefold. Firstly, to analyze the characteristics of labor force's structure, secondly, to compose unemployed profile and examine whether there is a change in the unemployed profile over the years and thirdly, to research the working conditions of employees and how they are employed in Turkey. In this thesis, Household Labour Force Survey (HLFS) micro data sets used which were derived from years 2014 to 2017. The labor structure of Turkey has been studied periodically and also discussed in a versatile mannerLogistic regression was used because it was suitable to work with categorical data. Decision trees, one of the classification algorithms, were applied to add more interpretation richness to the study. Turkey's labor structure is discussed under four chapters. To address the overall priorities of Turkey's labor situation "labor state" is analyzed. Unemployed persons, those in employment and those outside the labor force were handled together with multinominal logistic regression analysis. Then, the individuals in employment were discussed. It has been investigated whether the employees work in accordance with the maximum working hours of 45 hours per week determined by the labor law. The employees were analyzed by two class logistic regression whether they work according to the labor law or not. The characteristics and risk values of employees working more than 45 hours per week were calculated. Another analysis for the employees was based on the registration with the Social Security Institution and informal work. Groups at risk of informal employment were determined by two class logistic regression analysis. Unemployed individuals were dealt with decision trees according to job search periods. The period of job search is considered as unemployment for less than six months and for more than six months. The characteristics of short-term unemployed and chronic unemployed were determined. The unemployment period of the unemployed was examined according to the occupational group of the job sought.