Abstract
Predictive modeling is understanding complex sys- tems and decisionmaking informed by the parameter space of various domains. The aim of this study is to compare two predictive models, the Auto-Regressive Maximum Entropy Model (ARMEM) and the Decision Tree model, for predicting the performance of a specific variable, Surface Ocean Direction, from a dataset. The dataset, obtained through High-Frequency Radar (HFR) measurements around Koko Head, was taken as a case study to test these models. ARMEM is a time-series model that merges the autoregressive methods with the principle of maximum entropy, thus being highly appropriate for high-resolution spectral analysis and noisy or incomplete data. On the other hand, the Decision Tree model works through recursive partitioning of data and thereby provides intuitive, interpretable predictions through capturing the underlying linear and nonlinear relationships.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call