Lojistik regresyon ve karar ağacı algoritmalarının tahmin edici performanslarının karşılaştırılması: Yaşam memnuniyeti uygulaması

Decision tree algorithms and regression in machine learning create classes of data. Relationships between variables are modeled. Decision trees create classification rules using training data. They also test these rules on test data. Thus, the decision tree determines the success of the algorithm. With the model created in logistic regression, classification is created and classification performance is found. These methods are easy to interpret. They are easily applied to large data sets. They are used in many different fields due to the lack of assumptions. Satisfaction, which is a part of the concept of life satisfaction, is the fulfillment of needs, desires and wishes. Life satisfaction deals with a person's entire life. Life satisfaction is the whole of processes related to individuals' own life patterns and standards. The aim of this study is to compare the performances of logistic regression method and decision tree algorithms (CART, CHAID, QUEST) estimators using life satisfaction data (n = 8430) obtained by the Turkish Statistical Institute (TURKSTAT) for the year 2017. In this study, performance comparisons (accuracy, sensitivity, selectivity, precision, F-score) were made and it was found that the model that best explains the concept of life satisfaction is the QUEST algorithm.