A comparison of representations for the prediction of ground-level ozone concentration

This work presents a comparison of methods to predict ground-level ozone to highlight differences in the ability of the algorithms and to compare their performance to an established signal to noise based prediction method. Existing data related to weather conditions and ground-level ozone was divided into a training set and a test set. Three algorithms were trained using the training set to create predictors, which were then analyzed with the test set, and then compared to the Taguchi Method to determine performance. It was found that the newly introduced R-LCS performed well on this problem, predictors using the Taguchi method had a smaller deviation from actual results. This indicates an additional factor other than the level of correlation in the data that dictates how well these predictors perform on classification problems.