Student Dropout Prediction (Kalman Filter)

The Problem: STEM Retention High dropout rates in STEM fields are a persistent issue. Methodology: Psychometrics + State Estimation As a Research Assistant at the University of Tübingen Methods Center, I developed a prediction model supported by the Ministry of Education of Baden-Württemberg. • Kalman Filter: I applied multivariate time-series analysis (specifically Kalman Filters) to model student engagement as an evolving trajectory rather than a fixed point. • Latent Variable Integration: The model incorporated psychometric data, latent characteristics such as ability, motivation, and stress levels derived from longitudinal questionnaires. Impact The system functions as an early warning, identifying at-risk students weeks before they drop out. This enables educators to deploy timely, personalized interventions.