Using the Results of Capstone Analysis to Predict a Weather Outcome

Abstract

In this paper the results of capstone analysis is applied to predict a weather outcome using a decision-tree model. It examines weather data of the capital cities of Australia in a 12 month period to see if the decision-tree models can predict rain in Sydney the next day. It produces a decision-tree model with the raw weather data for each capital city to make the predictions about this outcome. It also aggregates the raw data to provide a combined-city dataset. Finally, it combines and compresses the raw data for each city using capstone modelling. The capstone data for the cities is used to train another decision-tree model to see if this provides better predictions of rain in Sydney than those obtained from using single-city models and the combined-city model. The results of these comparisons and details of how capstoning works are provided in the paper