2026/07/22 by Luis A Ladino, L. Ladino
Physics and Astronomy · Engineering · Social Sciences · #Experimental and Theoretical Physics Studies #Sports Dynamics and Biomechanics #Science Education and Pedagogy
paper · doi:10.1088/1361-6552/ae85a1
Abstract Experiments involving falling objects are fundamental in introductory physics to illustrate uniformly accelerated motion. However, traditional approaches often rely on idealized models that overlook real-world experimental variability. In this work, we present an educational framework in which students infer the drop height of a metallic ball from acoustic recordings of its successive impacts using a supervised random forest regression model. By extracting temporal descriptors from the audio signal, students analyse how physically meaningful features emerge from noisy experimental data. The model achieves high predictive accuracy ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:msup> <mml:mi>R</mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:mo>=</mml:mo> <mml:mn>0.996</mml:mn> </mml:mrow> </mml:math> ), but the primary pedagogical value lies in enabling students to identify which aspects of the physical process are most stable and informative. In particular, intermediate rebound intervals are found to be more reliable than the first impact, highlighting the limitations of traditional kinematic approaches. This activity integrates experimental physics, signal processing, and computational modelling within a Google Colab environment, promoting a transition from idealized problem-solving to data-driven scientific reasoning. The proposed framework provides a connection between classical mechanics and modern data science, preparing students to interpret complex experimental data in realistic settings.