Weather Forecasting: Using Time Series Analysis Under Different Stations Of The UK

Abstract

Time series is a statistical tool that is used for predicting future trends based on previous trends. In this study, forecasting of three weather stations of the UK presented. For that purpose, several methods are applied to the weather station data and the best fitting model is chosen for future forecasting. For the stationarity, informal (ACF and PACF) and formal (ADF) methods were presented. A well-known technique, Box-Jenkins (ARIMA), has been implemented. The evaluation for ARIMA (Auto Regressive Integrated Moving Average) model fitting and forecasting has been done through R software using various packages. Based on the inspection of the ACF, PACF autocorrelation plots, the most appropriate orders of the ARIMA models are determined and evaluated using the AIC-criterion.  In contrast to the respective models for the [1]station, ARIMA (2,0,1) and (0,1,1) for Cardiff and ARIMA (2,0,2) and (0,1,1) for Cambridge, respectively, are produced for the maximum and lowest temperatures at these stations. The annual as well as monthly analysis has been done for the validation of model. The result showed a good accordance of the projected temperature with real time data. Moreover, the ARCH/GARCH forecasting method was also presented on all three datasets. Additionally, the comparison of both ARIMA and ARCH/GARCH paradigms are presented. At the end of the study, drew a conclusion and discussed the further study gap as well.