Annual Rainfall Prediction Using Time Series Forecasting

Abstract

This paper attempts to determine which one of the various univariate forecasting techniques is producing accurate and statistically compelling forecasts for rainfall. The term “univariate time series” is referred as a time series that consists of a sequence of measurements of the same variable collected over regular time intervals. Forecasting techniques to predict rainfall are an important aspect as they are useful for business purposes, to take into account the transportation hazards that is a result of heavy rainfall, also it helps farmers and gardeners to plan for crop irrigation and protection. Most commonly, the techniques for prediction are regression analysis, clustering, autoregressive integrated moving average (ARIMA), error, trend, seasonality (ETS) and artificial neural network (ANN). In this paper, a review is provided based on different rainfall prediction techniques for predicting rainfall as early as possible. This paper has compared the performance of various forecasting techniques such as ARIMA, ANN, ETS based on accuracy measures like mean square error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE) and mean absolute error (MAE). On comparing these techniques, it is evident that ARIMA is performing well on the given data.