A Study of Time Series Models ARIMA and ETS
7 Pages Posted: 30 Jun 2017
Date Written: January 13, 2017
The aim of the study is to introduce some approach which might help in improving daily temperature of data. Weather is a natural a phenomenon for which forecasting is a great challenge today. Weather parameters such as Rainfall, Relative Humidity , Wind Speed, Air Temperature are highly non-linear and complex phenomena, which include mathematical simulation and modeling for its correct forecasting. Weather Forecasting is use to simplify the purpose of knowledge and tools that are used for the state of atmosphere at a given place. The prediction is becoming more complicated due to changing weather condition. There are different software and types are available for Time Series forecasting. Our aim is to analyze the parameters and do the comparison of some strategies in predicting these temperatures. Here we tend to analyze the data of given parameters and notice the prediction for few period using the strategy of Autoregressive Integrated Moving Average (ARIMA) and Exponential Smoothing (ETS).The data from meteorological centers are taken for comparison of methods using packages such as ggplot2, forecast, time Date in R and automatic prediction strategies are available within the package applied for modeling with ARIMA and ETS methods. On basis of accuracy we tend to attempt the simplest Methodology. Our model will compare on basis of MAE, MASE, MAPE AND RMSE. The identification of model will chromatic inspection of both the ACF and PACF to hypothesize many possible models will estimated by selection criteria AIC, AICc and BIC.
Keywords: ARIMA (Autoregressive Integrated Moving Average), ETS (Exponential Smoothing), AIC (Akaike’s Information Criteria), and BIC (Bayesian Information Criteria).
Suggested Citation: Suggested Citation