Abstract

Modified 2-Step Implicit Linear Multistep Technique for Time Series Smoothing

C. I. Nwokike1 , G. O. Nwafor2 , B. B. Alhaji3 , K. M. Koko4 , C. Nwutara5 , H. E Chukwuma6 , S. N. Nwanneako7 , O. G. Onukwube8 , J. O. Onyeukwu9

---

This paper presents a new time series smoothing technique developed from the modification of the 2-step implicit linear multistep numerical technique. The proposed time series smoothing technique was called 2-step mAMT in this paper. To test the efficiency of this new technique, it was applied to a real dataset (Nigeria external reserve, 1981 to 2015) with high noise (fluctuation). A comparison of the proposed 2-step mAMT was done with the simple moving average and simple exponential smoothing (? = 0.8). As shown in table 3, the proposed technique produced an ? 2 greater than the other techniques. The common indicators (measure of best fit) used are the MAE, MSE, RMSE, and MAPE as also displayed in table 3. Keywords: Time series, Linear Multistep, Adams-Moulton, Implicit, 2-step, Forecasting, Smoothing technique.

PDF