<![CDATA[Smoking is a behavior that is detrimental to all people. This study aims to determine the determinants of adolescent smoking. The data come from the March 2020 SUSENAS (215.679 teenagers). The model is composed of three binary logistic regression models, namely the model without resampling process, the model with random undersampling, and the model with random oversampling. The resampling technique was used because the number of teenagers who smoked was not balanced with those who did not smoke. The binary logistic regression model with resampling is the best model (86.54 percent balanced accuracy). The variables that affect the smoking status of adolescents are education, gender, marital status, occupation, and age. The type of ​​residence area also affects the smoking status of adolescents in the random oversampling model. Teenagers who tend to smoke are those who did not finish elementary school, male, married, work, live in rural areas, and older.]]>
Copyrights © 2022
