Result: PREDICTING MATHEMATICAL ANXIETY: A COMPARATIVE ANALYSIS OF RANDOM FOREST, LOGISTIC REGRESSION, AND HIERARCHICAL GENERALIZED LINEAR MODELS USING PYTHON

Math anxiety is a significant concern for many educators and policymakers due to its negative impact on students' math performance and career prospects. Various empirical studies have been conducted to examine the factors predicting math anxiety, typically based on a limited set of predefined variables, such as math performance and students' self-perception. However, to fully understand the nature of math anxiety, it may also be beneficial to conduct research based on more complex predictive models using a broader set of variables. In this context, our study aims to develop predictive models to forecast the level of math anxiety in students and identify key predictors influencing its occurrence. The study employs three statistical models: random forest, logistic regression, and hierarchical generalized linear models (HGLM). Each model has its own advantages and disadvantages. Applying all three models provides valuable insights into understanding the nature of math anxiety. The results show that the random forest and logistic regression models demonstrate high prediction accuracy. The key predictors identified by these models include self-confidence, problem-solving ability, and fear of failure. These factors significantly impact the development of math anxiety. The study's findings highlight the need for a comprehensive approach to identifying and reducing math anxiety. Practical recommendations include improving educational methodologies and supporting students' emotional well-being, which may help enhance their academic performance.