COMPARISON OF AUTOREGRESSIVE MODEL WITH MISSING DATA TREATED USING ORDINARY LEAST SQUARES AND INTERPOLATION WITH WEIGHTING METHOD

Abstract

Bandung is committed to contributing to the achievement of the Sustainable Development Goals (SDGs) in Indonesia. One of the efforts that can be made to support the 13th pillar of SDGS regarding climate change is to forecast the air temperature of Bandung City in the future. One of the models that can be used for forecasting air temperature data in Bandung is the Autoregressive (AR) model. Based on BMKG data, often the time series data obtained has missing data. Therefore, in order to do a good time series analysis, it is necessary to make an effort to correct the missing data. The purpose of this research was to examine the procedure for overcoming missing data in the AR model using the Ordinary Least Squares (OLS) method and Interpolation with Weighting, which was applied to forecasting the average air temperature data in the city of Bandung. The research methodology followed the Box-Jenkins 3-step procedure. The first-order AR estimation parameter model was estimated using the OLS method and then used to overcome missing data using both methods with weighting using R software. Both methods resulted in an estimated value of 0.9991 and the same Mean Average Percentage Error (MAPE) value of 2,459% with very accurate criteria. Therefore, to overcome the missing data on the average air temperature data in the city of Bandung with a parameter estimator close to one, we got the same result for both methods.