Authors
Arutyunov A.
|
Degree
|
Master of Technique, Volga-Dnepr Airlines LLC |
|---|---|
|
E-mail
|
artem.arutyunov@gmail.com |
|
Location
|
Moscow |
|
Articles
|
Aircraft market volume forecasting using Singular Spectrum Analysis and Survivor Curve Analysis
Market volume forecasting is one of the most essential steps in new aircraft development determining
critical inputs for sales, manufacturing, finance, after sales support, etc. For aircraft industry
forecasting challenges are due to technical characteristics, long-lasting development process and life
cycles, that requires forecasting for 10 – 20 or more years. Authors’ experience proves the importance
of combined analytical and expert approach in aircraft market volume forecasting for both short and
long-term periods. Analytical method is demonstrated in the article. Aircraft market volume is equal to
the sum of fleet development forecast and removal from service forecast minus current fleet in terms of
the number of aircraft. Fleet development forecasting is executed using Singular Spectrum Analysis.
Removal from service forecasting is executed with use of Survivor Curve Analysis. Both Singular
Spectrum Analysis and Survivor Curve Analysis are shortly described in the article. Results of passenger
widebody airplane segment forecasting for 20 years is given as an example. To test the efficiency
of the proposed analytical method forecasting using historical data is executed which means shortening
the initial time series on 5, 10, 15 and 20 points and executing a forecast on the respective time
period. Test results demonstrate precise fit of the historical data and forecast for 5 and 10 years forecasting.
For 15 years forecasting is properly done in terms of numerical aspect but improper in terms
of qualitative aspect. To test validity of the forecasting for 20 years the results were compared with
Boeing Current Market Outlook 2015, the difference was less than 0.3%. Reckoning on the Boeing’s
data is due to the hypothesis of lack of the statistical data while forecasting for 20 years.
Read more...
|
