Prediction of maximum 8-hour-average ozone concentrations using a generalized linear mixed effects model

摘要

The results of GLMM models show that these models can predict most ozone exceedance days correctly. In the validation stage, it shows that they are robust enough when they are applied to new data sets. Linear regression and partial linear regression which use same independent variables are compared with GLMM model. Results show that GLMM model can explain the most percent of ozone variance and its TPR in exceedance days is far beyond the other two models. Because GLMM model requires future meteorological information, its accuracy depends on the accuracy of weather forecast.