Abstract: In every manufacturing industry, forecasting is one of the most frequently applied analytical techniques that enables them to take very accurate and precise business decisions in day-to-day operations. In this regard, every organization is trying to set up the teams that should do one or other kind of predictions for their special needs and helping the organization to take data-based decisions. The key issue faced by many forecasters is accuracy of predictions made by them based on their available historical data. Many a times, the traditional approaches will fit the models based on generic trend and seasonal patterns but not specifically focusing on the trend of the previous value. In this paper, we tried to give an approach that tries to add a small correction factor to the predicted value in order to reduce the error in overall forecasting. This correction factor is based on past behavior on continuous trend in the historical data where the trend is just observed with respect to previous data. The correction factor on the either direction strategically reduces the amount of error in forecasting and also does not suggest any correction if it is already above the average positive or negative trend in the actual forecasted results. To illustrate this, we have taken the traditional forecasting methods like exponential smoothing and ARIMA and then applied the adjustment factor on the predictions from the above models. |
Keywords and phrases: time series forecasting, precision or accuracy, ARIMA, trend based correction factor.
Received: August 11, 2020; Accepted: November 27, 2020; Published: January 6, 2021
How to cite this article: P. Balasubramanyam and K. Sreenivasa Murthy, Trend based correction to the predicted results for time series forecasting, Advances and Applications in Statistics 66(1) (2021), 103-114. DOI: 10.17654/AS066010103
This Open Access Article is Licensed under Creative Commons Attribution 4.0 International License
References:
[1] P. Wessa, ARIMA Forecasting (v1.0.10) in Free Statistics Software (v1.2.1), Office for Research Development and Education, 2016. URL http://www.wessa.net/rwasp_arimaforecasting.wasp/. [2] R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 2nd ed., OTexts, Melbourne, Australia, 2018. OTexts.com/fpp2. Accessed on August 2, 2020. [3] E. Parzen, ARMA models for time-series analysis and forecasting, J. Forecast. 1 (1982), 67-82. [4] R. McCleary and R. Hay, Applied Time Series Analysis, Beverly Hills, CA, Sage Publication, 1980. [5] S. Makridakis and S. C. Wheel Wright, Forecasting, North Holland Publications, 1971. [6] N. Draper and H. Smith, Applied Regression Analysis, 2nd ed., Wiley, New York, 1981. [7] C. C. Holt, Forecasting seasonals and trends by exponentially weighted moving averages, International Journal of Forecasting 20(1) (2004), 5-10. https://doi.org/10.1016/j.ijforecast.2003.09.015. [8] P. Winters, Forecasting sales by exponentially weighted moving averages, Manage. Sci. 6 (1960), 324-342.
|