Forecasting

1
Road Map
 Role of Forecasting
 Forecasting Approaches

 Qualitative forecasting

 Quantitative forecasting

 Time Series Models

 Regression Methods

 Seasonal Adjustments

 Forecast Accuracy

Focus Forecasting 2
Forecasting
 Predicting the Future
 Vital for business
organization
 Underlying basis of
all business decisions
 Most techniques assume
an underlying stability in
the system
 Qualitative Forecasting
Approach:
 Quantitative Forecasting
Approach: 3
4
5
Qualitative Methods
 Grass root method – going down to the lowest level of
hierarchy
 Market research – data collection and hypothesis testing
 Jury of ‘executive opinion’ – source of internal qualitative
forecast
 Historical analogy – history or past data of the item
 Panel consensus – free open exchange in between select few
 Delphi Method - Iterative group process
 3 types of participants
 Decision makers: Evaluate responses and make decisions
 Staff: Administering survey
 Respondents: People who can make valuable judgments
6
Quantitative Forecasting
Time Series Models:
 Set of evenly spaced numerical data - Obtained by
observing response variable at regular time periods
 Forecast based only on past values - Assumes that factors
influencing past and present will continue influence in
future
1. Naive approach
2. Moving averages
3. Exponential smoothing
4. Trend projection

Associative Models / Causal Models:

5. Linear regression 7
Demand Behavior
 Trend
 Persistent, overall upward or downward pattern
 Changes due to population, technology, age, culture, etc.
 Cycle
 an up-&-down repetitive movement in demand over a
length of span
 due to business cycle; political and economic factors
 Seasonal pattern
 is often weather / festival / event / specific period related
 oscillating in nature - usually occurs within a single year
 Random variations
 Erratic; unsystematic; short duration non-repeating
 unpredictable and have no “assignable causes”
8
Forms of Forecast Movement
Demand

Demand
Random
movement

Time Time
(a) Trend (b) Cycle

Demand
Demand

Time Time
9
(c) Seasonal pattern (d) Trend with seasonal pattern
Components of Demand
Trend component
Seasonal peaks
Demand for product or service

Actual
demand

Average demand
over four years
Random
variation
| | | |
1 2 3 4
Year
Quantitative Forecasting
Time Series Models:
 Set of evenly spaced numerical data - Obtained by
observing response variable at regular time periods
 Forecast based only on past values - Assumes that factors
influencing past and present will continue influence in
future
1. Naive approach
2. Moving averages
3. Exponential smoothing
4. Trend projection

Associative Models / Causal Models:

5. Linear regression 11
Naïve Approach
ORDERS
MONTH PER MONTH FORECAST

Jan 120 -
Feb 90 120
Mar 100 90
Apr 75 100
May 110 75
June 50 110
July 75 50
Aug 130 75
Sept 110 130
Oct 90 110
Nov - 90 12
Simple Moving Average

n

i=1
Di
MAn =
n
where

n = number of
periods in the
moving average
Di = demand in period
i
13
3 Month Simple Moving Average

3
ORDERS MOVING
AVERAGE
 Di
i=1
MONTH PER MA3 =
Jan 120 – 3
MONTH
–
Feb 90 – 120 + 90 + 100
103.3 = 3
Mar 100 88.3
95.0
Apr 75 78.3 = 103.3 orders for Apr.
78.3
May 110 85.0
105.0
June 50 110.0
14
July 75
5 Month Simple Moving Average

ORDERS MOVING
AVERAGE 5
MONTH
Jan
PER
120 –
 Di
MONTH i=1
– MA5 =
Feb 90 –
5
–
90 + 110 + 130+75+50
Mar 100 – =
99.0
5
Apr 75 85.0
82.0 = 91 orders for Nov.
May 110 88.0
95.0
June 50 91.0
15
July 75
Weighted Moving Average
 Adjusts moving average method to more closely
reflect data fluctuations
 Weights are assigned to most recent data, barring in
case of seasonal cycles
 Precise weights are decided thorough trial and error
(based on experience and intuition), as does the
number of periods to be considered
 If recent periods are weighted too heavily, the forecast
might over-react to a random fluctuation in demand
 If they are weighted too lightly, the forecast might
under-react to actual changes in demand pattern
16
Weighted Moving Average

WMAn =  Wi Di
i=1
where
Wi = the weight for period i,
between 0 and 100 percent

 W = 1.00
i

17
Weighted Moving Average
MONTH WEIGHT DATA
August 17% 130
September 33% 110
October 50% 90
3
November Forecast WMA3 =  Wi Di
i=1

= (0.50) (90) + (0.33) (110) + (0.17) (130)

= 103.4 orders
18
Exponential Smoothing
 Averaging method - weights most recent data
more strongly
 As the past becomes more distant, the imp. of data
diminishes
 So very useful and preferable method, if recent
changes are significant and unpredictable
 Widely used, most popular because its an accurate
method
 Requires minimal data:
 forecast for the current period,
 actual demand for the current period and
 a weighing factor OR smoothing constant. 19
Exponential Smoothing
Ft+1 =  * Dt + (1 - ) * Ft
where:
Ft + 1 = forecast for next period
Dt = actual demand for present period
Ft = previously determined forecast for
present period
 = weighting factor, smoothing constant –
determines the level of smoothing
*Assume first forecast as Actual Demand…
20
Effect of Smoothing Constant
0.0  1.0
reflects the weight given to the most recent demand data
If = 0.20, then Ft + 1 = 0.20 * Dt + 0.8 * Ft
If = 0, then Ft + 1 = Ft
Forecast does not even consider recent actual data
If = 1, then Ft + 1 = 1 * Dt + 0 * Ft = Dt
Forecast based only on most recent data, so this becomes
as good as naïve forecast
21
Exponential Smoothing (α = 0.30)

PERIOD MONTH
DEMAND F2 =  D1 + (1- ) F1
= (0.30) 37 + (1- 0.3) 37
1 Jan 37
= 37

2 Feb 40 F3 =  D2 + (1- ) F2
= (0.30) 40 + (1- 0.3) 37
3 Mar 41
= 37.90

4 Apr 37 F13 =  D12 + (1- ) F12

= (0.30) 54 + (1- 0.3) 50.84
5 May 45
= 51.79

6 Jun 50
22
Exponential Smoothing
FORECAST, Ft + 1
PERIOD MONTH DEMAND ( = 0.3) ( = 0.5)
1 Jan 37 – –
2 Feb 40 37.00 37.00
3 Mar 41 37.90 38.50
4 Apr 37 38.83 39.75
5 May 45 38.28 38.37
6 Jun 50 40.29 41.68
7 Jul 43 43.20 45.84
8 Aug 47 43.14 44.42
9 Sep 56 44.30 45.71
10 Oct 52 47.81 50.85
11 Nov 55 49.06 51.42
12 Dec 54 50.84 53.21
23
13 Jan – 51.79 53.61
 Exponential Smoothing
70 –
 = 0.50
Actual
60 –

50 –

 = 0.30
Orders

40 –

30 –

20 –

10 –

| | | | | | | | | | | | |
0–
1 2 3 4 5 6 7 8 9 10 11 12 13
24
Month
 Regression Methods
 Linear Regression
 Regression can be defined as functional relationship
between two or more correlated variables
 Regression is used for forecasting by establishing a
mathematical relationship between two or more
variables (demand and some other independent
variable) in the form of a linear equation
 It is used to predict one variable given the other
 Linear regression refers to the special class of
regression where the relationship between the variable
forms a straight line
 Good for long range forecasting and aggregate 25
planning
 Linear Trend Line
Linear Regression is a causal
method of forecasting in which a
mathematical relationship is xy - nxy
developed between demand and b =
time. x2 - nx2
Linear trend line relates a a = y-bx
dependent variable (demand) to
an independent variable (time) in where
the form of a linear equation: n =number of periods
y = a + bx
a = intercept x
x= = mean of the x
b = slope of the line n
values
x = time period y
y = demand forecast for period x
n
y= = mean of the y 26

values
Least Squares Example
x (PERIOD) y (DEMAND) xy x2
1 37 37 1
2 40 80 4
3 41 123 9
4 37 148 16
5 45 225 25
6 50 300 36
7 43 301 49
8 47 376 64
9 56 504 81
10 52 520 100
11 55 605 121
12 54 648 144
78 557 3867 650 27
Least Squares Example
78
x = = 6.5
12
y = 557 = 46.42
12
b = xy - nxy = 3867 - (12)(6.5)(46.42) =1.72
x2 - nx2 650 - 12(6.5)2
a = y - bx
= 46.42 - (1.72)(6.5) = 35.2

28
 Linear Trend Line y = 35.2 + 1.72x
Forecast for Period 13 y = 35.2 + 1.72(13) = 57.56 units

70 –

60 – Actual

50 –
Demand

40 – Linear trend line

30 –

20 –
| | | | | | | | | | | | |
10 – 1 2 3 4 5 6 7 8 9 10 11 12 13
Period
29
0–
Linear Regression Example
x y
adv spend sales xy x2

4 36.3 145.2 16
6 40.1 240.6 36
6 41.2 247.2 36
8 53.0 424.0 64
6 44.0 264.0 36
7 45.6 319.2 49
5 39.0 195.0 25
7 47.5 332.5 49
49 346.7 2167.7 311
30
Linear Regression Example (cont.)
49
x= = 6.125
8
346.9
y= = 43.36
8

xy - nxy
b=
x2 - nx2
(2,167.7) - (8)(6.125)(43.36)
=
(311) - (8)(6.125)2
= 4.06

a = y - bx
= 43.36 - (4.06)(6.125)
= 18.46
31
Linear Regression Example (cont.)
Regression equation Sales forecast for 7 lakhs of ad spend
y = 18.46
60,000 – + 4.06x y = 18.46 + 4.06(7)
= 46.88, or 46,880
50,000 –

40,000 –
Attendance, y

30,000 –
Linear regression line,
20,000 –
y = 18.46 + 4.06x

10,000 –

| | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10
32
Wins, x
Correlation & Coefficient of Determination
 Correlation, r
 Correlation is a measure of the strength of the
relationship between independent and dependent
variables
 degree of association between two variables (-1.00 to
+1.00)
 nil/poor/average/strong, & positive/negative

 Coefficient of Determination, r2
 Percentage of variation in dependent variable resulting
from changes in the independent variable (0% to 100%)
 A measure of the amount of variation in the dependent
33
variable about its mean that is explained by the
Computing Correlation

n xy -  x y
r=
[n x2 - ( x)2] [n y2 - ( y)2]
(8)(2,167.7) - (49)(346.9)
r=
[(8)(311) - (49)2] [(8)(15,224.7) - (346.9)2]

r = 0.947

Coefficient of Determination
r2 = (0.947)2 = 0.897
34
Seasonal Adjustments
 Repetitive increase / decrease in demand
 Seasonal patterns can also occur on a periodic
basis
 Use seasonal factor to adjust forecast
 A seasonal factor is a numeric value that is
multiplied by the normal forecast to get a
seasonally adjusted forecast
 A seasonal factor range from 0 to 1, it is in
effect, the portion of annual demand assigned
to each season
 Thus SF when multiplied to annualDforecasted
i
Seasonal Factor =
demand yield seasonally adjusted forecasts for
each season S = D 35

i
Seasonal Adjustment (cont.)
DEMAND (1000’S PER QUARTER)
YEAR I II III IV Total
2002 12.6 8.6 6.3 17.5 45.0
2003 14.1 10.3 7.5 18.2 50.1
2004 15.3 10.6 8.1 19.6 53.6
Total 42.0 29.5 21.9 55.3 148.7

D1 42.0 D3 21.9
SI = = = 0.28 SIII = = = 0.15
D 148.7 D 148.7
D2 29.5 D4 55.3
SII = = = 0.20 SIV = = = 0.37
D 148.7 D 148.7 36
Seasonal Adjustment (cont.)
 X Y X*X X*Y
 1 45.0 1 45.00
 2 50.1 4 100.20
 3 53.6 9 160.80

 FIND MEAN OF X AND Y

 VALUE OF a AND b
37
Seasonal Adjustment (cont.)
For
2005
y = 40.97 + 4.30 x = 40.97 +
4.30(4) = 58.17
SFI = (SI) (F5) = (0.28)(58.17) = 16.28
SFII = (SII) (F5) = (0.20)(58.17) = 11.63
SFIII = (SIII) (F5) = (0.15)(58.17) = 8.73
SFIV = (SIV) (F5) = (0.37)(58.17) = 21.53 38
Forecast Accuracy
 A forecast is never ever accurate
 Large degree of error mean
 Either the forecasting technique used is applied
wrongly or is not applicable in the case
 Wrong relationship among variables
 Or the ‘parameters’ used need to be adjusted for
‘trend’
 Forecast Error
 Difference between forecast and actual demand - Error
 MAD - Mean Absolute Deviation
 MAPD - Mean Absolute Percent Deviation or MAPE
 Cumulative Error - RSFE
 Average Error or Bias 39
Mean Absolute Deviation (MAD)
S  Dt - Ft 
MAD= n
MAD: The absolute average difference between the
AD & FD.
where,
t = period number
Dt = demand in period t
Ft = forecast for period t
n = total number of periods
 = absolute value
40
The smaller the value of MAD relative to the
Other Accuracy Measures
 MAPD: Measures the absolute error (AV-FV) as a %
of demand rather than per period (MAD). Can be
used across the board to measure the relative
accuracy of the forecast.
 Cumulative Error (RSFE): Simply computed by
summing up the forecast errors. That’s why Linear
Trend Line has zero cumulative value.
 Average Error (Bias): Computed by averaging
the cumulative error value (RSFE) over the number
of time periods. +ve value: low, -ve value: high 41
and zero value: no bias
Other Accuracy Measures
Mean Absolute Percent Deviation (MAPD)
 |Dt - Ft|
MAPD =
Dt
Cumulative Error (RSFE)
RSFE =  et =  (Dt – Ft)
Average Error (Bias)
 et
E=
n
42
MAD Example
PERIOD DEMAND, Dt Ft ( =0.3) (Dt - Ft) |Dt - Ft|
1 37 37.00 – –
2 40 37.00 3.00 3.00
3 41  Dt - F3.10
S37.90 t 3.10
4 37 MAD = 38.83n -1.83 1.83
5 45 38.28 6.72 6.72
6 50 53.39 9.69
40.29 9.69
7 43 = 43.20 -0.20 0.20
8 47 11 3.86
43.14 3.86
9 56 44.30 11.70 11.70
10 52 = 4.85
47.81 4.19 4.19
11 55 49.06 5.94 5.94
12 54 50.84 3.15 3.15
557 49.31 53.39 43
Forecast Control
 Forecast can go out of control due to various reasons
 Change in trend
 Unanticipated appearance of a cycle
 Irregular variation such as unseasonable weather
 Promotional campaign, new competition, political
reasons, others…
 Tracking Signal: this indicates whether the forecast average
is keeping pace with any genuine upward or downward
changes in demand
 Monitors the forecast to see if it is biased high or low
 Control limits of 2 to 4(D
MADst -are
Ft)used most
RSFE frequently
Tracking Signal = =
MAD MAD
44
Tracking Signal Values
DEMAND FORECAST, ERROR E = TRACKING
PERIOD Dt Ft Dt - Ft (Dt - Ft) MAD SIGNAL

1 37 37.00 – – – –
2 40 37.00 3.00 3.00 3.00 1.00
3 41 37.90 3.10 6.10 3.05 2.00
4 37 Tracking
38.83 signal for period
-1.83 4.273 2.64 1.62
5 45 38.28 6.72 10.99 3.66 3.00
6 50 40.29 9.69
6.10 20.68 4.87 4.25
7 43 TS3 = -0.20 = 2.00
43.20 20.48 4.09 5.01
8 47 43.14 3.05
3.86 24.34 4.06 6.00
9 56 44.30 11.70 36.04 5.01 7.19
10 52 47.81 4.19 40.23 4.92 8.18
11 55 49.06 5.94 46.17 5.02 9.20
12 54 50.84 3.15 49.32 4.85 10.17
45
Tracking Signal Plot
3 –
Tracking signal (MAD)

2 – Exponential smoothing ( = 0.30)

1 –

0 –

-1 –

-2 – Linear trend line

-3 –
| | | | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10 11 12
Period
46
Example

Example 47
Example

48
 Forecasting Process
1. Identify the 2. Collect historical 3. Plot data and
purpose of forecast data identify patterns

6. Check forecast 5. Develop/compute 4. Select a forecast

accuracy with one forecast for period model that seems
or more measures of historical data appropriate for data

7.
Is accuracy No 8b. Select new
of forecast forecast model or
acceptable? adjust parameters
of existing model
Yes

9. Adjust forecast 10. Monitor results

8a. Forecast over
based on additional and measure
planning horizon
qualitative info’ & insight forecast accuracy

49