FORECASTING

I- What is Forecasting?
Forecasting is a technique to estimate, based on historical figures, expectations, trends, and/or
experience, a certain value of an uncontrollable variable for a certain future period of time.
Moreover, forecasting cannot be as simple as coming up with a figure, solely by considering
only historical data, without adjusting to other variables like competition, image and risk of
the country, interest rates, inflation, exchange rates, and other economical factors !herefore,
a person who forecasts shall adjust the forecasted figure to the realities and expectations for
the upcoming period in question
"n the scope of this very course, #ooms $ivision Managers forecast mainly #oom $emand
for a future period of time measured either in%
&umber of #ooms
&umber of #oom &ights
&umber of 'uest &ights
Room Nights ( )ccupancy #ate * +otel #ooms * ,verage -ength of .tay
Guest Nights ( )ccupancy #ate * +otel #ooms * ,verage 'uest per #oom
Forecasting demand in room nights and/or guest nights is a better measurement compared to
number of rooms. For, room nights and /or guest nights underlies more than one demand
dimension at the same time and hence is more meaningful
II- Forecasting Methods%
Forecasting
Judgmenta Statistica
!Time Series Modes"
Trend
#ro$ection
Trend %
Seasona
Smoothing E&'ert
O'inion
Mar(et
Sur)e*s
+e'hi
Techni,ue
/amel +otel0s +istorical #oom &ights for the past 12 years are depicted below%
-ear. Room Nights.
3454 6,712
3442 6,834
3443 8,262
3441 8,395
344: 8,631
3449 5,437
3447 7,9:2
3446 9,734
3448 7,214
3445 7,985
3444 6,864
1222 8,227
1223 8,:97
1221 5,:17
122: 5,487
1229 4,266
1227 4,317
1226 5,19:
1228 4,:19
1225 32,645
;ould you forecast 1224 room demand in room nights only bearing in mind the above<
mentioned historical data=
3. #ercentage Gro/th Method%
!he assumption underlying this method is that data in hand follow either an increasing or
decreasing trend !hat0s why> this very method shall be used, while forecasting, only when
data matches the assumption
$uring the last 12 years of operation of /amel +otel%
!he !otal ?ercentage ;hange in #oom &ights is @32,645 < 6712A / 6,712 * 322 ( 01234 5.
!he Bearly ?ercentage ;hange ( 69.25 C / 12 ( 6273 52
!he Forecasted #oom &ights for year 1224 is 32,645 * @3 D 2.2:12:A 889318 Room
Nights.
1. Mo)ing A)erage Method%
.imilar to the E?ercentage 'rowth MethodF, the Moving ,verage Method assumes an
increasing or decreasing trend
!his very forecasting technique aims at smoothing data and adjusts it as to minimiGe
volatility reflected in a high standard deviation between different records in the same data
range
!he most common used moving average is the $ouble moving average, which calculates a
third column by taking averages of couples of any two successive years -ater, the percentage
growth method would be applied to the smoothed data -astly, come up with the forecasted
value
-ear. Room Nights.
+ou:e Mo)ing A)erage
3454 6,712 <<<<<<<<<<<<<
3442 6,834 6,634.7
3443 8,262 6,554.7
3441 8,395 8,329
344: 8,631 8,:52
3449 5,437 5,16:.7
3447 7,9:2 8,381.7
3446 9,734 9,489.7
3448 7,214 9,889
3445 7,985 7,17:.7
3444 6,864 6,31:.7
1222 8,227 6,558
1223 8,:97 8,387
1221 5,:17 8,5:7
122: 5,487 5,672
1229 4,266 4,212.7
1227 4,317 4,247.7
1226 5,19: 5,659
1228 4,:19 5,85:.7
1225 32,645 32,233
!he !otal Moving ,verage ?ercentage ;hange ( @32,233<6,634.7A / 6,634.7 * 322 ( ;8276
5
!he ?eriod Moving ,verage ?ercentage ;hange ( 73.1: C / 34 ( 72<3 52
!he Forecasted 1225 < 1224 Moving ,verage ( 32,233 * @3.216467A ( 8397432=;1.
!he Forecasted 1224 #oom &ights ( @32,152.479 * 1A H @32,645A =9401 Room Nights.
:. Weighed A)erage Method%
!he Ieighed ,verage Method assigns ;ertain "mportance Factor or ;oefficient to
each historical Jalue. -ater, the forecasted value shall be computed by dividing the
weighted data to its coefficients by the sum of coefficients.
,ssigning weights or coefficients is and art that depends on experience, thorough
analysis of past figures, and performancesK Bet, whatsoever coefficients chosen,
there is always a certain subjectivity factor that might affect eventually the forecasted
figure
)ne of the most common types of the weighted average method is the simplest
method, which assigns the lowest weight to the oldest data in a sequential order.
!hough the simplest weighted average method is straight foreword, assigning least
weight to oldest data assumes that%
!he factors that affects the oldest demand diminishes through time and hence are
not important as far as the future period to be forecasted is concerned
!he factors and hence the conditions that created the last period0s demand are
assumed to continue heavily playing an important role in the next period to be
forecasted
.ince the above mentioned assumptions might not be valid, in most of the cases,
hotels shall adjust the coefficients attributed in the simplest weighted average method in a
way that mostly puts more weight on factors thought to affect next period0s demand and
less weight to those which would be considered relatively unimportant
,nother possibility is that hotels shall, after finding a forecast from the simple
weighted average method, adjust to experience, trends, and factsK
-ear Room Nights Weight Tota
3454 6,712 3 6,712
3442 6,834 1 3:,9:5
3443 8,262 : 13,352
3441 8,395 9 15,741
344: 8,631 7 :5,262
3449 5,437 6 7:,942
3447 7,9:2 8 :5,232
3446 9,734 5 :6,371
3448 7,214 4 97,163
3445 7,985 32 79,852
3444 6,864 33 89,974
1222 8,227 31 59,262
1223 8,:97 3: 47,957
1221 5,:17 39 336,772
122: 5,487 37 3:9,617
1229 4,266 36 397,276
1227 4,317 38 377,317
1226 5,19: 35 395,:89
1228 4,:19 34 388,376
1225 32,645 12 13:,462
Tota 783 890439666
!he 1224 Forecasted #oom &ights is 3,652,::: / 132 49337 Room Nights
9. Time Series Ana*sis%
!he !ime .eries Method tries through #egression ,nalysis to come up with a -ine that
minimiGes the distance between any ,ctual ?oint on the ;urve and its ;orresponding
?oint on the -ine @-east .quare MethodA. !his !echnique is refereed to as the #egression
,nalysis.
,fter finding the Lquation of the -ine @i.e. f @xA ( y ( a * x D bA, we try to forecast the
independent Jariable @in this ;ase, the 1229 Forecasted #oom &ightsA
,s far as /amel +otel0s problem is concerned, post to running a regression analysis, the
equation of the line that best fits the data @significant at a 47C interval levelA turned out
to be%
Room Nights > 80=260=7 ? -ear @ 668938=
!herefore the 1224 forecasted room nights ( 364.:641 * 1224 H ::3,234 =9711 Room
Nights
III- Com'arison oA Forecasting Methods%
Method. Forecasted Room Nights.
?ercentage 'rowth Method 889318
Moving ,verage Method =9401
.imple Ieighted ,verage Method 49337
#egression ,nalysis =9711
Managers shall use forecasting methods with extreme precautions, and shall consider first
the assumptions underlying each Forecasting Method to be able to find a good forecast.
,fter running a statistical forecast, managers shall adjust it to trends, expectations, and
opinion surveysK @".e. shall consider judgmental forecastingA