Simulating and Forecasting Periodic Behavior of a Discrete Time Series Using the Moving Average Method

Abstract

A new method for simulating and forecasting a discrete time series using the moving average method is proposed and substantiated. The idea of the method is to use the latus rectum as a quantitative estimator of the shape of a parabolic segment of a time series trend. Since the actual number of values of this indicator in terms of nature of its periodicity resembles an array of innovative buckets, there arose an idea of using the foci of parabolic segments which approximate these buckets. To build the forecasts, the simplest method of double moving averages of the third order was chosen. Using the moving average method, it is possible to obtain not only forecasts of points or intervals but also more complex ones, e.g., those of parabolic segments. The method of forecasting with the help of moving averages has provided an opportunity to give a generalizing picture of the changes in the process being studied, as demonstrated by the example of the indicator “Innovative Activity of Ukrainian Enterprises”. It has made it possible to predict the intensification of innovation in Ukraine for the period up to 2024.

Keywords

• moving average
• innovative activity
• double moving average