Forecasting of monthly temperature variations using random forests

This study utilized a random forest model for monthly temperature forecasting of KL by using historical time series data of (2000 to 2012). Random Forest is an ensemble learning method that generates many regression trees (CART) and aggregates their results. The model operates on patterns of the time series seasonal cycles which simplifies the forecasting problem especially when a time series exhibits nonstationarity, heteroscedasticity, trend and multiple seasonal cycles. The main advantages of the model are its ability to generalization, built-in cross-validation and low sensitivity to parameter values. As an illustration, the proposed forecasting model is applied to historical load data in Kuala Lumpur (2000 to 2012) and its performance is compared with some alternative models such as K-Nearest Neighbours , Least Medium square Regression , RBF (Radial Basic Function) network and MLP (Multilayer Perceptron) neural networks. Application examples confirm good properties of the model and its high accuracy