Xycoon logo
Time Series Analysis
Home    Site Map    Site Search    Free Online Software    
horizontal divider
vertical whitespace

Time Series Analysis - ARIMA models - Model Forecasting

[Home] [Up] [Identification] [ARIMA Estimation] [ARIMA Checking] [ARIMA Extensions] [ARIMA Forecasting]

V.I.4 Univariate Stochastic ARIMA Model Forecasting

The forecast (with forecast lead l) of a time series Yt at a forecast origin t is denoted by Ft+l or Ft(l) which is, in our case, always a combination of previous observations.

The forecast can obviously also be written in terms of previous error-shocks

Time Series Analysis - ARIMA models - Model Forecasting


Remark that from (I.III-17) it follows that the forecast variance is


and due to (V.I.1-238) and (V.I.1-233) it follows that




Hence, the forecast function (c.q. the conditional expectation of Yt+l at t) is the minimum mean square error forecast. Also it is quite evident that the forecasts are unbiased since


The variance of the forecast is equal to the first term of the RHS of (V.I.4-4). This illustrates why (V.I.1-51) (and especially the y-weights) can be used to compute the forecast confidence interval.

It is furthermore nice to see that the random innovations (c.q. shocks) are in fact equivalent to the one period ahead forecast error since


From (V.I.4-5) and (V.I.4-7) it follows that the one step ahead forecast errors are not correlated. Remark however that forecast errors for l > 1 are not uncorrelated.

There are two formulae for computing the correlations between forecast errors (l > 1): correlations between forecast errors at different origins with equal lead times, and correlations between forecast errors at equal origins with different lead times.

(Auto)correlations between forecast errors at different origins with equal leads are given by


which can be easily proved on using



which proves (V.I.4-8) (Q.E.D.).

(Auto)correlations between forecast errors at equal origins but with different leads are given by




Hence, the covariance is


which, together with (V.I.4-12), proves (V.I.4-11) (Q.E.D.).

Forecasting of ARIMA(p,d,q) models


is straightforward since


from which it follows that


Eq. (V.I.4-15) provides a recursive forecasting algorithm which can be solved using a computer and a programming language able to perform recursions.

The forecast variances (and hence also the confidence probability limits) can be obtained from the y-weights.

These y-weights are computed from


which can be used to equate coefficients of powers of B such that


Remark that if j > max(p + d - 1, q) eq. (V.I.4-18) reduces to


Sine the variance of the forecast is equal to the first term of the RHS of (V.I.4-4), it follows that


Confidence intervals can thus also be easily obtained.

Some explicit examples of the so-called eventual forecast functions are given in BOX and JENKINS (1976) and in MILLS (1990). These functions are used to study the fundamental nature of the forecasts. Since this goes beyond the scope of this work, and since the fundamental techniques of studying these forecast functions has already been given in previous sections, we only refer to the relevant literature.

vertical whitespace

ARIMA Estimation
ARIMA Checking
ARIMA Extensions
ARIMA Forecasting
horizontal divider
horizontal divider

© 2000-2022 All rights reserved. All Photographs (jpg files) are the property of Corel Corporation, Microsoft and their licensors. We acquired a non-transferable license to use these pictures in this website.
The free use of the scientific content in this website is granted for non commercial use only. In any case, the source (url) should always be clearly displayed. Under no circumstances are you allowed to reproduce, copy or redistribute the design, layout, or any content of this website (for commercial use) including any materials contained herein without the express written permission.

Information provided on this web site is provided "AS IS" without warranty of any kind, either express or implied, including, without limitation, warranties of merchantability, fitness for a particular purpose, and noninfringement. We use reasonable efforts to include accurate and timely information and periodically updates the information without notice. However, we make no warranties or representations as to the accuracy or completeness of such information, and it assumes no liability or responsibility for errors or omissions in the content of this web site. Your use of this web site is AT YOUR OWN RISK. Under no circumstances and under no legal theory shall we be liable to you or any other person for any direct, indirect, special, incidental, exemplary, or consequential damages arising from your access to, or use of, this web site.

Contributions and Scientific Research: Prof. Dr. E. Borghers, Prof. Dr. P. Wessa
Please, cite this website when used in publications: Xycoon (or Authors), Statistics - Econometrics - Forecasting (Title), Office for Research Development and Education (Publisher), http://www.xycoon.com/ (URL), (access or printout date).

Comments, Feedback, Bugs, Errors | Privacy Policy