Predicting Job Placement Anxiety and AI Dependency Classification Among College Students Using Random Forest

Abstract

The digital transformation driven by generative artificial intelligence (AI) has changed the landscape of higher education, but has also raised psychological challenges in the form of anxiety about the world of work and excessive reliance on AI. This study aims to predict students' job placement anxiety levels and classify their AI dependency levels using the Random Forest algorithm, while also generating personalised intervention recommendations. A secondary dataset was used, comprising 15,000 student records with 30 variables. The Random Forest Regressor model (150 trees, depth 15) produced an R² value of 0.815 for predicting job placement anxiety. Feature importance analysis revealed that interview anxiety (59.3%) and career clarity (11.8%) were the most dominant factors, while daily AI usage duration only contributed slightly (1.06%). The Random Forest Classifier model (150 trees, depth 15) achieved 72.0% accuracy in classifying AI dependency into low, moderate, and high categories. The recommendation system generated personalised interventions for 29.5% of students to reduce excessive AI usage. However, as a synthetic dataset was used, generalisation of the results to a real-life student population requires further validation. This study confirms that AI is not the main cause of career anxiety, but the replacement usage pattern needs to be managed with awareness and cognitive balance.