| Summary: | Accuracy in forecasting is vital to determine a precise result in every decision-making. However, massive spikes or outliers in a data series would interrupt the forecasting process. Hence, the smoothing method was introduced to prevent these factors from influencing forecast points. Throughout this study, 4253HT smoother was used to smooth a data series before Holt-Winter was applied in forecasting. Moreover, a comparison was made between the use of raw data and smoothed data in forecasting. The raw and smoothed data performance was evaluated using Residual Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and Akaike Information Criterion (AIC). As a result, the modelling 4253HT data yielded lower RMSE, MAPE and AIC values compared to raw data. The findings also presented that Holt-Winter performed better in forecasting with models that utilized 4253HT smoothed data. Furthermore, the application of 4253HT smoother on electricity data was made, where the outcomes indicated that it is a good model in data representation as the fitted line moves closer to 4253HT smoothed data. The use of smoothed values in forecasting has generated an excellent outcome as the data outliers are eliminated while its volatility is constant. © 2024 Author(s).
|
|---|