Abstract

Modified 3-step Implicit Linear Multistep Method for Time Series Forecasting

C. I. Nwokike1, G. O. Nwafor2, B. B. Alhaji3, K. M. Koko4, C. Nwutara5

---

Abstract In this paper, a time series smoothing technique derived from the modification of the 3-step implicit linear multistep numerical method was developed. The modified time series smoothing technique proposed in this paper was called 3-step mAMT and is applied to a dataset with a high fluctuation. The 3-step mAMT is compared with the simple moving average and the simple exponential smoothing (? = 0.8) and the 3-step mAMT performed better than the two other techniques. The measures of best fit used are the MAE, MSE, RMSE, and MAPE. The mAMT Order-4 produced a better result when compared to moving average and the simple exponential smoothing having an ?2 greater than the other techniques. Keywords: Time series, linear Multistep, Adams-Moulton, Implicit, 3-step, Smoothing technique, Forecasting.

PDF