..

ting & Facility Location (Module 3) 113

"' foreca8clical Variation:
''":. • CyeIical v~ati
· ·on arises for long time period. It
11) ~icallY bas four steps: upswmg or [Link]; the peak, at which
int activity levels out; the downturn; and the trough, or the lowest
Point For example, performance of the company for a fmancial year,
Poth enrolments in management colleges.
e . .
... ~ Flu~ tio~: The irre~ular fluc~ation is last component of
111~ tiIDC series.
:'filt~
uregu l~ fluf ~atio~s are not fixed and certain. Due to
Jllovem~nt m_·tune _~enes dt~: whic~ .~anno~ be explained by other
coJDPODents like ~ -~d or.~y~lical vana~ons ~~ irregular component is
ineluded -~o'~~~•~,
!;he ~gul ar .fl~ctua~on basically
· unus4al · events. ~uch _as -~ars, fl~s, hurricanes andoccurs after
political
disturbairce in a top leadership.
· ·· - t-cd ~d -, . . .- .- .
~¥~~ f§r_df; ~~ ~nd ~~ ~~ ~e~es ~~: , ._ .-. ~ _
Q SnnPle ~verage ~~,~-.~ -~~le averag~ -~~~~ ~ ~tennmed by:
. : .· . . : .. __ .:_ .Sumo f deQW1dsforallpastpen09S .•
SunplcAverage-
· · · ::. Numuc
.•·--,.'i...- fd
ro emand peno
.., . . , d .
s ·"
'
. · _· . . · .- o..+ ti2 .+ I>3 ~ ......+ Dn · ,s. ,: .~J ·
• . = _ . ; ; . _ . . . . ; ; ; _ _ . : a , __ _~ .
... - .....
• •

. , .- .·: n ~ :· - : ,. . , - ~- .'!,,
,
wlieie~ t>1 i-aemand ofperiocl'N~.-i: i.e:·:th~"mosf~ece~t-perloci .
• , ..,_ t .:: - ... . ,.. • ...•
.-02 =demand .of period No:·2, .i.e_.~.~e ~nod .jus~-prior to [Link] l '· ~ f :
4'o ., ·' • " • • • ·• ,'' 1 -:.,. • .,. • '" . ~. • 't• '-,

. ..._pe~ .. _;· ---,.;_: ... ·, . :·: ,.· ·:: . :;•·•.·:· ,. ! ~'• ·. ..· .
~=delhand'ofperioo No. 3, i.e., the period prior to ~riod No. 2
n. =·deiiisod of period N,o.'n . ,. . . .. .. · -"--~ . :~- . ..: ._ -
. . - . -.· .. . ); .. . . •. .,. . . ~ . : ·t 1} .,. , ' ' . '. ,; •

The objective of simple_~yerage meth~ .is to e s ~ .th~-. pa~m or

central tendency of demand. In simple average method, .the deinand is
equally distributed in all the periods. If there. high cleman~ in some are
period, it will probably compensate by'fow 'demand in th~ otlier periods.
The c~ces of influence from random fluctu¢on_is minimised by the
help of averaging which may occur in any single period. The basic
disadva
I •
ntage of this method is that it uses the data of the demand which is
• •
~ t recent leaving the data of demand which .has happe~ed in past. It
may lead inaccurate data for deciding future course of action. There· could
also be some advancement in the technology or some changes in the work
method which may not due to their weight on the results. . •.. • .
, Example 1: The demand for 80 Watt bulbs in the months of January,
February, March and April was 300, 400, 600 and 500. Forecast the -
monthly demand for the bulbs. ·
Soluti~n: Simple Average.(SA) ~ Di + D 2 + D3 + D4• ·
.n . . . ~
~

~ 300 + 400 + 600 + 500 = 1,800 ·=·450 .

'Tb '. '. , . . . '
. .; . 4- ' . . 4 .
e average de~d for 80 Watt bulbs is 450 per month.
116 . . ·. ·
· MBA Third Sem
ester (Projects and
· · · · . Working Operations Man~..
notes: · ,~
- 1st moying tota
l= 7 + 6.7 +7.3
= 21 .
: / :··. , nd m~vin
2 g total= 6.7 +7.3
. .. ·. :: . . · · . + 7.7 = 20.7, and so
21 on
• .... 1~t m~v~g .
.. ' ~verage = = 7
. .'...- ...~ -- · _- ·. - _. ·· ·.· .. 3
2nd.m~~n~ avera 20.7
ge= - - = 6 .9
3 and so on
· -~-- , b) . Even
- P~riod. of M o ~ g Aver
· .·._.· :·. ·. averag age: If the perio
e is ,4, -~' or _8, _i d of
· · · cannot b t 1s an even number
e placed agamst . ~ e four-yeany
.· the second .and any year as the
the third year. S median 2.S is
.·· between .th~ 2 o , the total shou
nd ~ d 3rd yea ld be p l~·
.•
.
.. .. . . . . · av~ra ge in ?rder to pla rs . W e m ust centre the mo·
ce the moving av
erage against the
The steps are: year.
• Compute the
. ~ ·in between values of the first four years an
the [Link] the 3 d place the taa
- · • Leave th rd years.
e first·year
, ;· four years·and pla value and compute the value of
ce the total in bet thei
· ' ! . ·. · . , · •
This process mu ween the 3rd and r.a
·· ·. · st be continued 4th yea.
into account. until the last year
is tam
• Compu~ the
first tw
middle year (i.e., o four-year totals and place it agai
3rd year) . nst Ill
• Leave the firs
t four y
year totals and pla ear total and compute th~ next tw
ce in the 4th year. o fOIII·
• This method
must be continued
are computed. until all the four-y
early touh
• Divide the ab
ove to
four-yearly totals tals by 8 (because it is the total of
) and put in the the Iii
trend value next colum.n. Thi. 15. twO
. . .·· s
'
Example 4: (Fo
r even number
data, calculate 4 of items): -From wing
year moving aver the folio
Year . 1
age.
Outout . 40
2
42
3
46
4
44
s
so
6 I1 Is
I9
48 I 54 I 52 I
10
56 60
Solution: Cal~
Year Output ulatlon o f 4 earl es of the Data
4 Year Mo\'ln& e
4 Year
(1) (1)
Total . Movln& Total or Column Year Centerd
(3) . (4) ·
Moving A,era,
Averaae · (6 ) • (5) +2
1 4 = 3 +4 5
40
2 42
3 172 43
46 . 88.50 ·
44.25
18 2 [Link]
4 .. '46.25 .
' .
. M Forecasting & Facility Loca~ion (Module 3) 117
pO' ' •.· . 188. 47 .
' 50 . .·. 96 . • 48
.5 '49 : . ·
48 ' · · 100 · . 50 -1
.6. .. . ·. ' 204 .. : !.-· · · · • 51 .
. '
54 .~ ~ · . \ . ,: · ,: 103.~0• · · 51.75
7 ·. 210 -<·. · 52.50 · ·
·, . 54
.8 52 108 .!
.222
·. 56 . .....
.
'
9

. 10 60 '
:. . . ~-- ~. < ~. .~ :-~-\ .'_ . .-. ~-· ,. . . . .~t:_.:·.. ·.. •.• .. ' . '
' ••• :; ._J•• •, -: • • . : · '. • • , : :· \

_g Notes
I ,; ,
Workin
. .. : ·. -·. . ' a , •• • • " • ·• , • ·: ; •
1st 4~Year Moving Total =40 :+- 42 + 46-+ 44 =172 . . ~

· 2nd 4-Year Movmg_Total =42 + 46 + 44 ·+50 = 182 · -· ..' ·.. . · · .: ,

• • • _,. I •' I •• - • I \ ,r • • " ·,, • ,t l

1st 4-Year Moving''Average '~ 172''f 4 = 43, ~d·so on ·..,- . .. ·

1st 2 Itein Mov1~g T<?~ = 43 :t:.45-?0; ~8.5,o,:and so ori ' - .
. ·.1st [Link] ~~ier_~d)\1~ying Av~Jag~.= ~~-_50 + 2 ~ 44._25 .
iv) ~Weighted. ~~~:,:A~erage. M~tho~: --~ ·#s ·, ~etho4, :_the ·trend _
. analysis·-is ~alculate4; by .multiply~g 1th~ -·moying totals by weights
assigned to them ~d ·the _weighted .average is calculated by dividing
the product by the:sum of the weights. . ' l ' ••. · .. · · .
~ ... C f .. • '•. • • : • • •• ' •• I ..

.Example 5: Find the trend for th~·follow~g se~es usi~g a three-ye~

weighted movin avera e with wei ht 1, 2 and 1.
·· Year · · : 1· 2 3 .4 5, 6 7 .
Value : 6 12 15 21 24 - 30 39
Solution: The weights of the three years are respectively 1, 2 and 1
+
and their sum is 1 + 2 1 = 4. . ·
, Weighted M~ving Average=
·
~y:, :Ew :4 : · ,
Com utation of 3~ earl Wei ted Movin Avera e
Year Values 3-Yearly Weighted Moving Totals 3-Yearly Weighted
, Y · Yw MovlngAvenge ·
(1) (2) (3) 3 + l:w
1 6 . (
- .
2 . , 1~ (lx6) + (2xl2) + (lxl5) = 45 45 +4 = 11.25
3 15 (lx12) + (2x15) + (lx21) = 63 . ~3 -:t- 4 = 15.75
4· 21 (lx15) + (2x21) + (lx24) = 81 81+4=20.25
5 · 24 (1><21) + (2x24) + (lx30) = 99 · 99+4 = 24.75
6 30 (lx24) + (2x30) + (lx39) = 123 ' 123 + 4 ~ .75.
7 39 ,. ..
v) Exponential Smoothlng: Exponential smoothing is. a very common
method and. is also useful in operation management. The basic reason
for acceptance are as follow: _ ; .· · · -
. a) They are immediately available in· standard ·computer software
packages. · ·. · ·· . ··
b) . They basically need less data [Link] calculations. '
 MBA Third Semester. (Projects and Operations ~
118 . ethod the past data of demand is u anagen...
Under this m ' ed h SCd c ·"1111, ting & Facility Location (Module 3) 119
. . h The weights us are owever exponential . 0 llsi~.
we~g ts. allotted to the most recent period and it 10 na111~·,, pOtd forccas
' _ Actual demand (or sales) for the previous period (say July)
·ally as we go down. It can also be said th &oes on ~-
weights areu IL, D1-1 -
exponen . at the ..~ . F = 500 + 0.2 (450- 500)
are-• l·near rather they are exponential and the ~g "'-°CiDi.. ~ ,
•• I : 500 + 0.2 (-50)

. hts decrease for later years. n11u~. = 500 - 0.2 X 50

we1g • "I:°'\ = 500- 10 = 490 units
Uses of Exponential Smoothi~g I :. Sales forecast for August is 490 units.
a) There arc correct exponentJ~ models. . ,
b) The preparation of exponential model 1s easy. I Double-Exponential Sm~thing . .
c) The working of the model c~ be_und~rstood by the u j Double-exponential smoo~g refers _to an exponential s~oo~g over an
d) There is less work of ~culatlon m this model. sers. I already smoothed time senes. It penruts the data forecasting with trends. It
e) There are less requirements of computer Stora • manages only the random trends. It handles the procedure of expon~ntial
minimum use of past data. ge as Uicit smoothing to a time series which is an exponentially smoothed [Link]
f) The accuracy test can be easily calculatep. 1 give detail of linear trend in the forecasted value. The extrapolated senes
is characterized by the constant growth rate, equal to the growth of
Single Exponential Smoothing smoothed series at the end of data period. The upward or downward trend
following arc two requirements for creating the eq . in collection of data over a series of one period results in delay in
update the forecast: uation of ~
1 exponential forecast. It can be more or less the actual occurrence. The
a) The existing demand for the most recent time period correction in the exponentially smoothened forecast can be done by
b) The cuncnt demand forecast · adding the trend adjusonent while the correction in trend can be done by
As each time period expires, a new forecast is made· smoothing the constants. The trend equation also takes the help of
1
Forecastof next period's demand · ' • , . constant delta. If in case the alpha and delta is not used then the trend will
lead to the enors. [Link] helps in minimizing the effect of the error
=a(Actualdemand ~or )+(I-ex)(Demand forecast for) which may arise between actual and forecast.
:. /
most recent penod most recent period The trend equation is manually entered in initial stage to let the trend
F1 =aD 1_ 1 +(1-a)f;_ 1 , equation going. It can be a prediction or can be calculated from past data.
where, 0 SoS l, and tis the period The equation to compute the Forecast Including Trend (Fil) is
1
f1T1 = F1 + T1 ......(1)
If the time period t-1 gets over then the actual demand D t . .1
t - l. In the startin of •od . . 1-1 or pcooj1 F1 = FlT,_1 + cx(A i--1- FIT 1-1) .....(2)
demand d . .g pen t-1, if there is forecast FL1 ofi T, ~ T 1-1 + cx6 (A ...,...: FIT ,_1)
should be
fi ,L.
'::g pcnod ~ -1. Hence, at the finishing of t - I 11111
th the requuement for computing the forecast of ~111111·
where,
.....(3)

or we next period FL F1 = The exponentially smoothed forecast for period t

T, = The exponentially smoothed trend for period t
Example 6: ABC company sed . : FIT1 = The forecast including trend for period t
method . u simple exponential sm~
to f~°:e
was 500 units
a;:ncntial smoothing constant of 0.2 (i.e., OJ
·term demand. The forecast for the monlh of 111
a= FIT1-1 · = The forecast including trend made for the prior period
A 1-1 = The actual demand for the prior period
a = Smoothing constant
the forecast i ' wthhercas tbe actual sales were only 450 units Whal i
or e month of August? · ' In case the trend is already there then the second smoothing applied is
Solution: F, =aD,_1 +(1-a)F,_1 I constant. For making a trend-adjusted forecast, following process is done:
or, F, = F1-1 + JX (D1-I - F1-I ) a) The single exponential forecast is made with the help of previous
where forecast, actual data and smoothing constant with the help of:
F, = Fi-1 + a(A1-1 - F1_1)
F • = Forecast for the cu . b) Compute the trend:
F._, = Forecast the . rrent penod (say August)
a. = smoothin previous period (say July) T, = TH + Jl(F, - F1-1)
8 constant (a= 0.2) c) The trend needs to be added to the forecast:
TAF=F,+T
 MBA Third Se~ter (Projects and Operations M
. EJample ?: In the last pcnod. we had a trend-adjusted t ~~ I
120 121
ting & Facility Location (Module 3)
the acniaI demand was 4~0, an~ we used 0.2 for bo o , \ p0fd, ~ t of Correlati on (r): The coefficient of correlation 'r' shows the
<l ~ Oft,
mponent in !he pnor pcnod was found to be th8Ca)
trend co . P. ~ coefflc ~:e importance of relationship between x and y. The absolute
for caste. cu~ ·, 1
compar; shows the kind of relation between x and y whereas the sign of r
O
value t: the positivity or negativity of the relationship. The symbol of
Solution repr~;. of regression line should be same as the symbol of coefficient of
a) Fi = 400 + 0.2(420 - 400) = 404
b) T= 8 +0.2(404-400) = 8.8 ~::Jation . Toe c~fficien t of correlation (r) is determined by the help of
c) TAF = 404 + 8.8 = 412.8 the following equation:
nl:xy-l :xLy
TAF here is the trend-a(ljusted forecast. For the 0
t,ecomes F.., and 8.8 becoF Ts-1, ~= ✓[n1:x2-(I:x)2) [n1:y2 -(Ly)2]
1
. ext PcriO(I, 4
1

• ·Th~ value of r is always betw~en HI ~d (+)l. ·The ~alue of y incre3;Ses if

Triple-Eiponelldll Smoothing th value of r is + I in proportton and v,ce versa., So, 1{ the value of r 1s -1,
Sometimes ill noo-linear trends, it is unportant to . '
exponmtia1 smoothing. It i s ~ to manage the parabc) ~ the~ . it ~e~cri~s ~ perf~~ neg,a~ve rela~onship b~t if~ in~eases th~ value _o_f y
trends '-l i·l .._.,, decr~li. ~d )li_Ce ver~p., If, r JS +-0.3_ It. w,!I show_ we3:1' pos1?ve
usually used fir the dataThe [Link] exponential smoothin
' :' rel11tionstpp. If r is·,-:a.$,.•_t shows the strong !legatlve relationship. If r IS 0
mlic: siluatioo in demand parameters and doublc-exi>Otlc .g , •,
, then there,~-no relatJon&hip ~e~ween X and y. ' . . ' ' '> .
will ~ in tbe ~ 0 0 in which demand is changing ~tial -~
and [Link] manages all the other business time ~ ~--• if
;·:·: ~~~~lions' Re~'iof An~ysi s '· '·: . .' : • ..
2) [Link] RqadllW W This is one of the most cornrrton1y 'Setit.s. · I The [Link]·of
[Link] uld be
regression
relationsh
analysis
ip
are
between
as
the
follows:·
dependen [Link]
' .
independent
in ~ of demand. Here. relation between quantity Used~ .-..i):·
• .....,, variable and otbec indcpcDdent variables like pric.e ~ ~ .·., variables.. , ·: · · . •, · ! · ·, '. •· ' · · ·;

substimrcs, de. is established and a regression equation is ro:mU::, ~- •. ' ii) . The regression .analysis only calculates the values which are within the
, range not for the [Link] the range. ,· ·
+ bX. Such rdalioo between the dependent and the ~ ~ Yt1
iii) , It is·assumed•that the relationship between the dependent and independent
often awar one but soo:Jdimes it may also be curvy-linear After ~ i
~- .,.:., .. ,variables is [Link] till the regression equation is calculated.
t h e ~ equalioo, tbe value of variable y can b e ~ f ~
by IISiiig• , iv)· The random· value can bl: ,taken for dependent variable but for the
value of .~bt variable X and the two constants ·
, . , independent .variable, the fixed value is·taken without any error and
~ anal~ ims ~ from the least-square techni i.e. ; and b. 1't
, that is chosen by the analyst or the user. ·
to odn' economic. competlttve, or internal variables
~=:ootheis ~which ~the re':3tionship. The ~hnique is
oquatton
As :

describes the relationshi betw

anal~
Y:~ .to Iii
~
ecn two er 1111
v) , Only .one · dependent variable is required for dependent variable
, ,, , . , :•whereas more than one independent variable can be used. ·
iodq,mdeot [Link] · le . . P
framed on ...... .:ght WJC:
"-- ~ regression analystS, a regression eqtlalioii E~ple 8:_ Sale of flats in one of the cities during the past 8 years has
-. as given below: been mcreasml? as can be seen from the followin2 sales data:
Y=a+bX I 2 3 4 5 6 7 8
Year
where, ~~:odent variabl~ shown on the Y-axis of the graph.
• ' . dependent vanable shown on X-axis of the graph
Sale of Flats
(Nos.) .
45 48 44 46 49 51 50 53

8
, , = mterccpt on the Y-axis, and ·
b = slope of regression line. Determine the following:
i) What is the forecasted sale of the flats in the next two years? Use
The following equati • regression analysis for detennining the forecasted sales.
on gives the value of constants 'a' and 'b':
2 ii) Calculate Coefficient of Correlation. Interpret its meaning. .
a= LX LY-rxt xy
b= ntxy- txtxy iii) Calculate Coefficient of Determination of the sales. Interpret its meaning.
nrx2 -(1:x)2 iv) Draw the data of sale on a graph. Draw the regression line on the graph.
ntx 2 -(tx) 2
Forecast the sales of the tlats in the next two years from the graph.
The dependent variable y i
and 'b' and the inde iic:rec~t can be_ calculated by the values ii~· Solution:
upon future value. pen t Variable X m regression equation ~ i) Process the data for regression analysis to forecast the sale of flats.
Details arc shown in table below.
 S mester (Projects and Operations M
MBA Thi·rc1 e
122
vear Sale ofFlats (Nos.) 1 ..: xy ~y2 ~ ), . g & Facility Location (Module 3)
M foiecasun .
123

...-I
X
45 . 1
4
45
96 ~
pO .'. coefficient of Correlauon
u) nLxy-Lx LY
48 2,304
2 9 132
3
44 1,936 , =-{nLx2 -(Lx)2](n L y2-(Ly)2J
46 16 184
4 2,116
49 25 245 8(1,782) ·-36(386)
s 51 36 306
2-401
2 2
6 2,601 =- fs<204)- (36) ](8(18,692) - (386) ]
7
so . 49 350
2•500
53 .• 64 424 14,256 - I 3,896
~
8 I

ts= 1:J:386 l:'l:'=204 LXY= 1,1s2 tr =- Jti,632 - l,296)(1,4 9,536 -1,48 ,996)
:::: 18,692
36

As it is known !hat,
---= : 360
== Jj36 x 540
=
360 360
=-=0.845
./1,81 ,440 426

rx2 :Ey-l:xl:xy _ 204(386) - 36(1,782) '

1 coefficient of correlation between time period and number of sale of
o:Ex2 -(I:x)2 - 8(204) - (36) 2
-
flats is 0.845. The- strength of relationship is represented by the
- 78,744 -64,152 = 14,592=43.43 absolute value of coefficient of correlation (r = 0.845). [Link] shows a
3 36 strong relationship between the time period and number of sale of
- 1,632 -1,296
flats. If the value of coefficient of correlation is zero, it shows no
b = n l: xy - Ix Ly = 8(1,782) - 36(386) relationship whereas if the absolute value is +I, it shows an absolute
n l: x 2 - (L x) 2 8(204) - (36) 2 relationship. As the direc tion of relationship shown by the value of r is
positive (+),'the sale of flats increases as the time increases.
=14,256 -13,896 =360 =l.071 iii) Coefficient of Determination (r2) = (0.845)2 =0 .71 or 71 %.
1,632 -1,296 336
A coefficient of detennination of 71 % shows that 71 % of the variation
Therefore, y = a+ bx= 43.43 + 1.071x . in y or sale of flats is explained by the variation in x or the time
ss period. The balance 29 % variation in sale of flats is due to other
S4 Y,n=54.14sav S4, - factors.
..
S:l Yo - 53.07 say 53 _ _......- -
I
E SI
52
.,,,.,,..
i..----- -
8 iv) The grapb between time (years) and the number of sale of flats is
•shown in figure 3.4. The sale figures are ·plotted on the graph.
::, 0
- 'Based on the value of •a' and 'b' calculated above, the regression
~ so ,,,,...
~
0
line AB is drawn. From the regression line the forecasted sale of
!!
• 49 ~" flats is given below:
...
Ii: 48
0
I) 47
-; 46
(I)
,,,,
./"
'
Regression Line AB
y9 (sale of flats in year 9) = 53 flats.
• Y10 (sale of flats in year 10) = 54 flats.
45 Y - 86.57 + 2.095X
44 .-:::.- - }•= 86.571 lntH>'• ~ on '1-axis) 3) Trend Projection Method: The data of the past can be used as a stepping
I 2 3 4 s 6 7 ll 9 I II stone for the future demand projections. A trend can be established on the
basis of the past figures. S uch trend can be graphically or equationally
1 Years established. The equation method is the accurate. It can be calculated by
' various methods which are as follows:
Figure J.4: Sale of Flats In Various Yeanr
If forecast for the sale oftl 10. .1k • . . Freehand Method (Graphical Method)
Y10 then: ats the 9 year is y9 and in the 1~ year• It is one of the most easy, simple and adaptable method used for
Y9 = 43.43 + 1.07 I (9) _ 4 43 c~culating trend. The process used for calculating the straight-line trend by
- 3. + 9.64 = 53.07 say 53 flal8. as
this method is follows·
Y10 = 43.43 + l.07I (JO)- · ·
- 43.43 + 10.71 = 54.14 say 54 flats. i) The lime series on the gruph can be plotted.
- ~ l
 :s.....~----=--~~ -
~~~~-

~
,m.A 1bird Semester (Projects and Operat·
l\'l" ionsM
.
124
V
'-i
in~g
.ung & Facility LocaUon
i:ore'115
(Module 3) 125
,ofd, method shows the trend value of the overall time period. For the
ii) 'fhe i·ng average method, there is no trend value for the first few and
'
movJast few 11me pen'ods. For exampIe, in. the movmg . average trend
: \ years, there will be_ no trend v~ue fo~ firs! two and last two y~ars.
... BY the help of equation of str~ght hne, 1t shows the functional
-+x 111) Jationship between X and Y senes and can also be used to forecast
Figure 3.S: Freehand Method :ture values. It is not applicable in moving average method.
. In the case of linear trend, the technique shows the growth rate per
ii) Study carefully the direction of the trend based 0
iv) annum for year data. If in the equation y = a + bt, the value of t has
information (dots). n the
t,een calculated by the change i~ the origin as D = 0 then the number
iii) Toe straight line can be_ made which suits the da of year is odd for yearly data b 1s the annual rate of growth and when
personal judgment The line reflects the direction of ta ~
the lrcn~"III& the number of year is even then 2b is annual growth rate.
Toe trend line by the free-band method uses the follow·1ng po·. · ExaJII le 9: The annual sales of com an 'Dev' are ·ven below:
1
accurate trend line: n1s to Year · 2000 2001 2002 2003 2004
i) Toe free band curve which is made should be smo th KIi
Sa)es In Lacs {f) 90 112 156 92 150
ii) Generally, there should be same number of poin: 3•
the curve. . . bov~lQd ~ BY the method of least square find the trend value for each of the five
iii) There should be same total dev1auon above the tr d .-
. en line as ~
I years-A)so estimate the annual sales for 2005.
the vertical deviation below the line.
iv) Toe total of squares of the vertical deviation from the · •\ Solution: Let the straight-line trend be represented by the equation y = a +
very small. trend ~J bt. The values of a and b shall be determined by solving the following
nonnal equations:
Importantt oC Freehand Method . l:y=na+b:Et .....( ])
i) Freehand method is used in every business as ·t
1 · . l:yt .=aD + bll .. ...(2)
saves labour and time. is simple and,
ii) It is more adjustable than the mathematical function S0 . .
curve accurately to the data.
I
· , it adJusts•
Herc, since the number of years is odd the mid-year, i.e., year 2002 is taken
as origin and one year as unit.
iii) It is more used in applied statistics as the stati
accurate line than a mathematician.
ti .
s c1an c~ draw lllt
I Computation of Trend Valoes of Sales
Time Deviations Square of Product of Time
Years Sales in
iv) It assists in making clear the character of ti
suitable mathematical trend can be used.
.
me senes and 1111
I Lac(~ from the Year Time Deviations (t) and
(y) 2002 (t) Deviations Sales (y)

4) ~ Square Method: This method assumes consum tio

I 2000 -2
(tl)
4 -180
previous rate of change continues to prevail in futu A p . n _Iha! * 90
mathematically conside · th re. trend line 1s dmi 2001 112 · -1 l -112
between calculated d nng e past data in a manner that the diffCl!II 2002 156 0 0 0
formed which is an that of observed value is minimum. A trend liK1; 2003 92 l 1 92
. the best fit for the available data The tr d Ii . ......'
on two variables where one variabl 15 . . en nc is ... 2004 150 2 4 300
independent Th; val . e dependent and the other vari&Nii
the value of the de ue;:_f tbe •_odcpend~nt variable is put forth to estiml
n=S .ty= 600 l:t=O :Er-10 l:yt= 100
I
cost associated wi:~ methnt van~le. This method is simple to use amil, :. n=5
od lS also very less.
Putting the values in the nonnal equations
g'1pc,~:~Leaa Square Method
with the fr== depends on ~bjective, and biasness
judgement in the co curv~. nus helps in reducing the ~
is not
·
p«d,
LY=na+b1)
=> 600 = 5a + (b x O)
=> 600 = Sa
•

mputatton. => a•l20

 ~~;";;,'-. -:. .-,:::;_~ ~ ~ - -
:sr;::,a;::i,;:.:i;:;<

•"'-A Third Semester (Projects and OJ>erai·IOflsi.

, 127
1n.u 'lity 1.,oeation (Module 3)
126 "I . & Fac1
.J>l.4 r,orecasung a}ue of 'b' in equation (1), we get
1 -a1)+bl:i2 fll'.., tting the v
:~oo::(a><O)+(b>< IO) 13~~.82
~ tOO== !Ob ,~ al of 'd' and 'b' in the regression equation of 'Y' on 'X':
~ b::10 '·' ttiJ)g the v ue
uation is y = 120 + IOt with origin at 2nn,,. ', BY P11 + bX, we get .
:. 'Ibe treDd eq VVl 1111d
~~:;1:82 + 3.97 X
year. ' I~
a)ue oft for 2005 will be 3. Hence the estimate for th • .. . ted sales for the year 2015:
1bc ;JOOS= !20 + (10 x 3) = 120 + 30 = 150 .~Y~21lJi, fs~ 37,82 + 3.97 X 3 .
~f'' :: 37.82+ ll,?l :49.73
Exam le 10: K.G. & Co. has conducted investigation re .
ceJn commodi and collected the followin datai &ard~g that for the year 2016:
3
Year 2009 2010 2011 2012 2013 ' and 6:: 37.82+ 3.97 x 4
2 YJJI :: 37,82 + 15.88 = 53.70
25 30 .--.1.._;_:_.
35 014
40l.._:38 .. 47
Sales (in • ~.---.L..--.L.
~~~:..:.;!..J •c Model: The econometric model forecasting involves
forecast the sales for the year 20 I5 and 20 I 6. . , . .. S) ~no~eu:everal simultaneous equations, which are, generally, behavioural
,
• • J ~.
~ g IJlllthematical identities and market-clearing equations. The
Solution: . . . equauonstn,•c model technique is also known as simultaneous equations
Com utation or Trend Values or Sales or a Certain Com · . . , ecoooroe " . This teehniquc uses
Yem Sales (in T'une Square of Time p~ IIIClhod and complete s~stem app~a~h to !orecastlllg.
'000 Deviations Deviations De • Uct oft._ . u·cated 1Jlllthemat1cal and stat1st1cal tools. ·
VJ~Ualons {X)III saphis ' ·
Unit) from the (Xl)
(Y) Year 20U • ,, ts!Y) rtance of Econometric Method ·
~'Ibis method possess~ all the a~vantages of th~ regression meth~.
200') 2S -3 . .. . '9 ' . . • I 1 • -is . ") purthennore, if one 1s not womed about the size of the model, 1t can
20!0 30 · -2 . , .- .4 •.• , · . _,, ,:•. , ~ , u a]mOst eliminate the regression method's major problem of forecasting
2011 35 -1 I · • · -35 ·the values of the independent variables in the prediction period.
2012 40 o . • o_ 0 iii} Provided the data are available, one should not really be concerned
2013 38 1 I
2 4
. 38 with the model's size, for the computer is there to help the forecasters
2014 47 94
IX 2: 19 to estimate and solve the model.
N=6 IY=21S IX=-3
Regressioo equation of 'Y' on 'X'.: Ye= a+ t bX
• .# ,_
• I , ...•.:. i ., , · • • ' Limitations of Econometric Method
i) To the extent the structural changes have taken place, the past
To ~md out the value of 'a' and 'b', ·th~ follo~ing two [Link]
' 1.,
applied: · • relationship between independent and dependent variables will not
l:Y =Na+ bI:X continue in the future.
ii) The forecasts under this method assume that the forecast equation
I:XY =atx-b tx2 holds exactly in the prediction period. To the extent it is stochastic,
i.e., the disturbance tenn is non-zero, the forecasts wm be wrong.
By putting the available values in the above equations we get
215=6a-3b •
.....(1)
-38=-3a+ 19b
.,, ..(2) 3.4. FACILIT Y PLANNI NG
By multiplying equati00 (2) by 2 and adding it with equation (1), we"' Facilities can be at the macro level as well as at the micro level. The macro
,..
21 S= 6a- 3b
_ 76 = --6a + b : 1._facilities_ may comprise of warehouses, plants, stores, cross-docking
38 liues, tcnrunals, and customer sites, whereas the micro level facilities arc
139 = 35b r,l'of larger_ or macro level facilities. For example, a factory (macro level
b =3.97 :~) may mclu?e a machine shop, paint shop, ll!ld a welding shop; and out
e, the machine shop may include milling machines, grinding machines,