Keeping the Same Rule
Forecasting is not an easy task; we have to agree in what path we have to move ahead,
There is one trail forecasting approach, but if we try get the same answer for all our business units, that’s will be great, I know it is a tough way, but it can be achieved in these days.
We force all the three seasonality models results to follow the trend one
Yes we keep the same rule but in one step ahead – it is in the FUTURE
The CMO Survey - Highlights and Insights Report - Spring 2024
Keeping the Same Rule
1. Keeping
The Same Rule
By: Mohammed Salem Awad
www.slideshare.net/airports_forecasting
2. ON TARGET
Data Input
“Excellence is never an
accident.
It is always the result of high
intention, sincere effort, and
intelligent execution; it
represents the wise choice of
many alternatives - choice, not
chance, determines your
destiny.”
―
Aristotle
3. ON TARGET
Outline
Keeping The Same Rules
Forecasting Approach
Input Data
Objective
Seasonality Model
Seasonality Model
Final Results
Forecasting Accuracy Matrix
Conclusions
Trend Forecasting
Seasonality Model
Contact
6. ON TARGET
Forecasting Approach
We will use the concept of
Forecasting by Objectives
to develop a fair matrix decision, so
forecasting by objective ; can be either by:
- Classical Method by Evaluation R2
- Setting Signal Tracking S. T. (36 ) to Zero
- Defining the Max/Min S. T. in the control
band.
- Targeting the final results of the annual long
term forecast.
- Reflecting the impact of the most recent
monthly data.
7. ON TARGET
Forecasting Approach
Defining the Max/Min S. T. in the control band.
Golden Rule -4 < Signal Tracking < + 4
And Coefficient of Determination > 80 %
10. ON TARGET
Trend Forecasting
Input Data :
Based on 21 data set (21 years - from 1992- 2012). By implement trend approach
using the best of line fit ( Power Function ) the results of fair fitting are
R2 = 96.5 while Signal Tracking = ± 5.71
The Forecasting of 2014
= 54,203,771 Pax
Max/Min Signal Tracking Analysis:
The aim of this analysis is to keep
most of the signal tracking values in
constrain band ( -4 and + 4 )
maintaining high value of R2 .
The graph shows the residual values
by yellow color are out of the band
for 21 set data base, which reached
the highest extreme value by ± 5.71.
11. ON TARGET
Trend Forecasting
R2 = 96.5 while Signal Tracking = ± 5.71
The Forecasting of 2014 = 54,203,771 Pax
12. ON TARGET
Seasonality Model ( Short Term ) :
Europe + Intercontinental = x
Generally speaking the normal
method to evaluate short range
data with seasonality impacts is
AREMA Model, but in this
analysis we will try use the best
of art technique that reflect two
parameters only, they are
displacement and Rotational.
Our approach is to find the line of fit that passing through the year
of accumulated forecasted figures of 12 months for 2014, and that
reflects a minimum errors and high relation factor ( R2 ) for both
series ( Europe & Intercontinental ) which satisfies the following
relation
Europe + Intercontinental = x
17. ON TARGET
Seasonality Model ( Short Term ) :
O & D + Transfer = x
Generally speaking the normal
method to evaluate short range
data with seasonality impacts is
AREMA Model, but in this
analysis we will try use the best
of art technique that reflect two
parameters only, they are
displacement and Rotational.
Our approach is to find the line of fit that passing through the year
of accumulated forecasted figures of 12 months for 2014, and that
reflects a minimum errors and high relation factor ( R2 ) for both
series ( O & D and Transfer ) which satisfies the following relation
O & D + Transfer = x
22. ON TARGET
Seasonality Model ( Short Term ) :
Scheduled + Unscheduled = x
Generally speaking the normal
method to evaluate short range
data with seasonality impacts is
AREMA Model, but in this
analysis we will try use the best
of art technique that reflect two
parameters only, they are
displacement and Rotational.
Our approach is to find the line of fit that passing through the year of
accumulated forecasted figures of 12 months for 2014, and that
reflects a minimum errors and high relation factor ( R2 ) for both
series ( Scheduled + Unscheduled ) which satisfies the following
relation
Scheduled + Unscheduled = x
29. ON TARGET
Forecasting Accuracy Matrix:
Forecasting Accuracy Matrix can be represented by four regions i.e Fair ,
Mislead, Poor, and Unrelated, for our cases : only one case (Transfer) is FAIR
as it is satisfied the pre- request
constrains while most of the other
segments are Mislead which actually fairs
results that deny the mislead issue for the
following reasons :
- The Signal Tracking
values are defined on
both sides of the trend
line so the issue of
displacement is not
exist.
- By visual inspection,
the forecasted model is
lay on the actual data.
31. ON TARGET
Conclusions:
The study shows, that there is possibility to design our targets even
though to have same target, off course it hard task but it needs
patience and time to deliver a fine results.
The rule of the signal tracking is to refine the final results and
positioning the trend line in the final direction of analysis.
Two methods can be used to get the forecasted figure of 2014 =
= 54,203,771 Passengers either in one step ( analysis ) based on 72
data set – optimum case which is applied.
Or in two steps ( two analysis ) one optimum and the other one is
adjusted based on 36 data set each.
All data segment are reported, and any researcher can compare the
forecasted figure by the actual data to evaluate the forecasting
approach.
The study shows high accuracy.
33. ON TARGET
Contact :
Mohammed Salem Awad
Consultant
Email:
smartdecision2002@yahoo.com
www.slideshare.net/airports_forecasting
Tel: 00967736255814
P.O. Box: 6002
Kahormaksar
Aden
Yemen
Date of Issue: 07 MAR 2014