Independent Modelling Investigation
Bivariate (2D) & Multivariate (3D) Mathematical Modelling
Abstract
This investigation develops a mathematical model of projectile accuracy by examining the individual and combined effects of throwing distance and consecutive throw number. Motivated by the possibility that accuracy changes through initial technique calibration and subsequent fatigue, the experiment considers how these effects interact across different distances.
A dataset of 400 observations was collected by recording the radial deviation of consecutively thrown marbles from a target at eight distances between 0.5 m and 4.0 m. The resulting data were analysed through separate univariate models for distance and throw number before being integrated into a bivariate model describing their combined effects. Logistic, quartic and piecewise functions were explored, with root mean squared error (RMSE) used to assess model performance.
The analysis identified a sigmoidal relationship between distance and radial deviation, alongside three distinct regimes in consecutive throwing: an initial calibration phase, a relatively stable period of optimal accuracy, and a subsequent deterioration attributed to fatigue. These relationships were combined into a three-dimensional model that captured the interaction between distance and throw number, achieving an RMSE of 4.154 cm. The investigation demonstrates how multivariable mathematical modelling can be used to represent complex real-world behaviour while critically assessing the limitations of model fit and extrapolation.