Alternative Method: Outlier Treatments with Box-Jenkins and Neural Network via Interpolation Method
AbstractOutliers represent the points that greatly diverge and act differently from the rest of the points. These kinds of phenomenon usually happen in the data especially in time series data. The presence of this outlier gave bad effect in all statistical method including forecasting if there are no actions on it. Thus, this paper discusses alternative methods which are linear interpolation and cubic spline interpolation to the time series data as outlier treatment. Assuming outlier as missing value in the data, the outlier were detected and the results were compared using forecast accuracies by two popular forecasting model, Box-Jenkins and neural network. The monthly time series data of Malaysia tourist arrival were used in this paper from 1998 until 2015. The result indicates that the improved time series data using the linear interpolation and cubic spline interpolation showed great performance in forecasting than the original data series.
Open access licenses
Open Access is by licensing the content with a Creative Commons (CC) license.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.