FORECASTING

FTN3OB3

MR S. SITHOLE
September 2024
 Outline
What is Forecasting?

Forecasting Time Horizons

Forecasting Approaches – Qualitative and

Quantitative

Qualitative Forecasting

Quantitative Forecasting

2
 Specific Learning Outcomes
The student should be able to:

Understand what is forecasting.

Understand the three time horizons and which

models apply for each.

Explain use of qualitative models.

Apply quantitative forecasting methods.

3
 What is Forecasting?
The art and science of predicting future events.

May involve: ??
- future projection of historical data
using a mathematical model.

- subjective or an intuitive prediction.

- future projection of demand-driven data.

- combination.

4
 Product Life Cycle and Forecasting
Introduction – Growth – Maturity – Decline

 Product’s position on its life cycle may influence

forecast.

 Introduction and growth require longer forecasts

than maturity and decline.

 As product passes through life cycle, forecasts are

useful in projecting – Staffing levels; Inventory
levels; Factory capacity.
5
 Importance of Accurate Forecasting
 Supply-Chain Management: Good supplier
relations, advantages in product innovation, cost
and speed to market depend on accurate forecasts.
 Human Resources: Hiring, training, and laying off
workers depend on anticipated demand.

 Capacity: Capacity shortages can result in

undependable delivery, loss of customers, loss of
market share.

6
 Forecasting in Disney Land
 Forecasting provides competitive advantage.

 Forecast attendance and behaviour at each ride.

- Staff of 35 analysts and 70 field people survey

one million park guests, employees and travel
professionals each year.

- Generates daily, weekly,

monthly, annual, and
5-year forecasts.
7
 Forecasting in Disney Land
 Forecast used by labor management, maintenance,
operations, finance, and park scheduling.

 Forecast used to adjust opening times, rides,

shows, staffing levels, and guests admitted.

 Disney characters are used to entertain guest when

lines are forecasted to be long.
On slow days, fewer cast
members are called to work.

8
 Forecasting Time Horizons
1. Short-range: operational (< 3 months, up to 1 year)

- used to plan purchasing, job scheduling, workforce

levels, job assignments, and production levels.

2. Medium-range: tactical (3 months to 3 years)

- used in sales, production planning, and budgeting.

3. Long-range: strategic (more than 3 years)

- used in planning for new products, facility location,

research and development.

9
 Comparing Time Horizon Differences
 Short-range forecasting tend to be more accurate
and usually employs different methodologies
(mathematical techniques) compared to long-range
forecasting.

 Medium and long-range forecasts deal with more

comprehensive issues and support management
decisions regarding planning and products, plants
and processes.

10
 Types of Forecasts
1. Economic forecasts (medium to long-range)
- Address business cycle by predicting inflation
rate, money supply, housing starts, etc.

2. Technological forecasts (long-range)

- Predict rate of technological progress.
- Impacts development of new products.

3. Demand forecasts
- Predict sales of existing products and services.
11
 Common Assumptions of Forecasts
 An assumption is made that what happened in
the past will happen in the future.

 Forecasts are never perfect.

 Forecasts for groups of items are more reliable

than forecasts for a single item.

 Long-term forecasts become unreliable.

12
 Characteristics of Good Forecasts
 Accurate – small forecast errors (deviation from actual
demand).

 Consistent – reliable and consistent in its ability to track

actual demand.

 Timely – available within a reasonable time frame to

make necessary decisions.

 Simple – easy for users to interpret.

 Efficient – cost of preparing forecast should not

outweigh its benefits.
13
 Steps in Forecasting

1 • Determine the purpose of the forecast.

2 • Select the items to be forecasted.

3 • Determine the correct time horizon.

4 • Select the correct technique/ model.

5 • Gather and analyse the data.

6 • Make the forecast.

7 • Validate and implement the results.

14
 Forecasting Approaches

Qualitative Quantitative

Jury of Executive Opinion Time-Series

Naive Approach
Models

Delphi Method Moving Averages

Sales Force Composite Exponential Smoothing

Market Survey Trend Projection

Associative
Linear Regression
Models

15
 I. Qualitative Methods
 Involve decision maker’s intuition, emotions,
subjective interpretation, personal experience, and
value system.

 Used when situation is vague and no measurable,

reliable, historical or statistical data exist.

- New products

- New technology

 Four qualitative techniques.

16
 1. Jury of Executive Opinion
 Uses the opinion of a small group of high-level
managers (operations, finance, marketing) to form a
group estimate of future demand.

 Advantage – combines managerial knowledge and

experience with statistical models; relatively quick.

 Disadvantage – domineering
person can impose his/ her
decision to other members.

17
 2. Delphi Method
 Uses a group process that allows experts (5 – 10) to
make forecasts. Decision Makers
(Evaluate responses
and make decisions)
 3 types of participants
Staff
(Administering
- Staff personnel survey)

- Respondents (outside)
Respondents
- Decision makers (People who can make
valuable judgments)

 Iterative group process, continues until consensus is

reached.
18
 3. Sales Force Composite
 Based on salespersons’ estimates of expected sales.

- Each salesperson estimates sales in the region.

- Estimates from individual salespersons are reviewed

to ensure that they are realistic.

- Combined at district and national levels to reach an

overall forecast.

 Sales representatives know customers’ wants.

 May be overly optimistic.

19
 4. Market Survey
 Customers or potential customers are asked about
future purchasing plans.

 Useful for demand and product design & planning.

 Uses information not available anywhere else.

 May be overly optimistic

- What consumers say, and what they actually do

may be different.

20
 II. Quantitative Methods
 Employ mathematical modeling to forecast
demand.

 Used when situation is stable and historical data

exist.

- Existing products

- Current technology

 Two categories: Time-series and Associative.

21
 A. Time-Series Forecasting
 Time-Series models uses a series of past data
points (week, month, quarter, annual) to make a
forecast.

- Breaks down past data into components and

project them forward.

- Four forecasting methods.

 Time series has four components: Trend,

Seasonality, Cycles and Random variations.
22
 Time-Series Components
1. Trend – gradual upward / downward movement of
data over several years (long
term) due to changes in income,
age, population, etc.

2. Seasonality – data pattern of up

and down fluctuations occurring
within a single year due to weather,
customs, etc.

23
 Time-Series Components
3. Cycles – repeating up / down
movements of multiple years
due to business cycle, economic,
0 5 10 15 20

political factors.

4. Random variations –
unsystematic fluctuation caused
by unforeseen event, which is of
short duration and non-repeating.

24
 1. Naive Approach
 Short range forecasting technique.

 Assumes that demand in the next period is equal to

demand in the most recent period.

- If January sales were 68, then February sales will

be 68.

 Sometimes cost effective and efficient.

 Can be a good starting point against which more

sophisticated models that follow can be compared.
25
 2. Moving Averages
 Short range forecasting technique.

 Uses a number of historical actual data values to

generate a forecast.

 Smooths out short term irregularities in data series.

 Simple moving average - used for little or no trend.

 Weighted moving average – detectable trend.

26
 Simple Moving Averages
Month Actual Sales 3-Month Moving Average
January 10
February 12
March 13
April 16 (10 + 12 + 13) / 3 = 11 2/3
May 19 (12 + 13 + 16) / 3 = 13 2/3
June 23 (13 + 16 + 19) / 3 = 16
July 26 (16 + 19 + 23) / 3 = 19 1/3
August 30 (19 + 23 + 26) / 3 = 22 2/3
September 28 (23 + 26 + 30) / 3 = 26 1/3
October 18 (29 + 30 + 28) / 3 = 28
November 16 (30 + 28 + 18) / 3 = 25 1/3
December 14 (28 + 18 + 16) / 3 = 20 2/3

27
 Weighted Moving Averages
Weights Applied Period  More responsive to changes
3 Last month
- more emphasis (weights)
2 Two months ago
1 Three months ago on recent values.
6 Sum of the weights
Forecast for this month =
 Weights are assigned
3 x Sales last mo. + 2 x Sales 2 mos. ago + 1 x Sales 3 mos. ago based on experience and
Sum of the weights
intuition.

Month Actual Sales 3-Month Weighted Moving Average

January 10
February 12
March 13
April 16 [(3 x 13) + (2 x 12) + (10)] / 6 = 12 1/6
May 19 [(3 x 16) + (2 x 13) + (12)] / 6 = 14 1/3
June 23 [(3 x 19) + (2 x 16) + (13)] / 6 = 17
July 26 [(3 x 23) + (2 x 19) + (16)] / 6 = 20 ½
August 30 [(3 x 26) + (2 x 23) + (19)] / 6 = 23 5/6
September 28 [(3 x 30) + (2 x 26) + (23)] / 6 = 27 ½
October 18 [(3 x 28) + (2 x 30) + (26)] / 6 = 28 1/3
November 16 [(3 x 18) + (2 x 28) + (30)] / 6 = 23 1/3
December 14 [(3 x 16) + (2 x 18) + (28)] / 6 = 18 2/3

28
 Graph of Moving Averages

Weighted moving average

30 –

25 –
Sales demand

20 –

15 – Actual sales

10 – Moving average

5–
| | | | | | | | | | | |

J F M A M J J A S O N D
Month

29
 Problems with Moving Averages
 Requires extensive historical data.

 Lag the actual values – does not forecast trends

well.

 Increasing n smooths the fluctuation better but it

makes the method less sensitive to changes in
data.

30
 Moving Averages Exercise
 Daily high temperature in Johannesburg for the last
week are as following:
Days Sun Mon Tue Wed Thu Fri Sat
Temperature (°C) 21 22 18 20 21 24 23

(i) Determine daily high temperature for next Wed

through Saturday using a 3-day moving average.

(ii) Determine daily high temperature for next Wed

through Saturday using a 3-day weighted moving
average with weights of 0.2, 0.4 and 0.6.
31
 3. Exponential Smoothing
 Short range forecasting technique.

 A form of weighted moving-average method where

data points are weighted by an exponential function.

- Gives weight to all past observations.

- Most recent data weighted high and have more

influence.

 Involves very little record keeping of past data, and

fairly easy to use.
32
 Exponential Smoothing
 Basic formula:

New forecast = Last period’s forecast + a (Last

period’s actual demand – Last period’s forecast)

 Mathematically: Ft = Ft – 1 + a (At – 1 - Ft – 1)

where Ft = new forecast

Ft – 1 = previous period’s forecast
a = smoothing or weighting constant
(0 ≤ a ≤ 1)
At – 1 = previous period’s actual demand
33
 Exponential Smoothing - Example
 Predicted demand for February = 142 bread loaves.

Actual demand for February = 153 bread loaves.

Smoothing constant chosen by management, a = .20

 Determine new demand for March.

New forecast for March demand (Ft)

= Ft – 1 + a (At – 1 - Ft – 1)

= 142 + .2 (153 – 142)

= 142 + 2.2 = 144.2 ≈ 144 loaves.

34
 Exponential Smoothing - Exercise
 Daily high temperature in Johannesburg for the last
week are as following:
Days Sun Mon Tue Wed Thu Fri Sat
Temperature (°C) 21 22 18 20 21 24 23

Determine daily high temperature for next Monday

through Saturday using exponential smoothing with
a smoothing constant of 0.5. The forecasted daily
high temperature for next Sunday is 26 ℃.

35
 Exponential Smoothing with Trend
Adjustment
 When a trend is present, exponential smoothing must
be modified.
Month Actual Demand Forecast (Ft) for Months 1 – 5
1 100 Ft = 100 (given)

2 200 Ft = F1 + a (A1 – F1) = 100 + .4 (100 – 100) = 100

3 300 Ft = F2 + a (A2 – F2) = 100 + .4 (200 – 100) = 140

4 400 Ft = F3 + a (A3 – F3) = 140 + .4 (300 – 140) = 204

5 500 Ft = F4 + a (A4 – F4) = 204 + .4 (400 – 204) = 282

Forecast estimate: 1st month – perfect.

2nd, 3rd, 4th and 5th months – severely lagging.
36
 Exponential Smoothing with Trend
Adjustment
Forecast including Exponentially smoothed Exponentially smoothed
= +
trend (FITt) Forecast average (Ft) trend (Tt)

Ft = a (At - 1) + (1 - a) (Ft - 1 + Tt - 1)

Tt = b (Ft - Ft - 1) + (1 - b)Tt - 1

where Ft = exponentially smoothed forecast average in period t

where
where Tt = exponentially smoothed trend in period t
At A = actual
t -=1 actual demand last period
demand
- 1t = actual demand last period
at =
Ft -F -=1 smoothing
= forecast
forecast last
last periodfor average (0 ≤ a ≤ 1)
constant
period
b
T
1
= smoothing
= trend constant
estimate last for trend (0 ≤ b ≤ 1)
period
Tt - 1 = trend estimate last period
t - 1

a =a smoothing
= smoothing constant
constant forfor average
average ≤ a≤ ≤a 1)
(0 (0 ≤ 1)
b =b smoothing
= smoothing constant
constant forfor trend
trend ≤ b≤≤b1)
(0 (0 ≤ 1)
37
 Trend Adjusted Exponential Smoothing -
Example
 Forecast demand for an equipment using:

Increasing trend is present.

Smoothing constants: α = .2 and β = .4

Month 1: Initial forecast average (F1) = 11 & Trend (T1) = 2

Month (t) Actual Demand (At) Month (t) Actual Demand (At)
1 12 6 21
2 17 7 31
3 20 8 28
4 19 9 36
5 24 10 ?

38
 Trend Adjusted Exponential Smoothing -
Example F = a (A ) + (1 - a) (F t t-1 t-1 + T t - 1)

Tt = b (Ft - Ft - 1) + (1 - b)Tt - 1

Actual Smoothed Forecast Step 1: Average for Month 2

Smoothed
Month (t) Demand Forecast Including
Trend (Tt)
(At) Average (Ft) Trend (FITt) F2 = a A1 + (1 – a)(F1 + T1)
1 12 11 2 13.00 F2 = (.2)(12) + (1 – .2)(11 + 2)
2 17 12.80 1.92 14.72 = 2.4 + (.8)(13) = 12.8 units

3 20 15.18 2.10 17.28 Step 2: Trend for Month 2

4 19 17.82 2.32 20.14 T2 = b (F2 - F1) + (1 - b)T1
5 24 19.91 2.23 22.14 T2 = (.4)(12.8 - 11) + (1 - .4)(2)
6 21 22.51 2.38 24.89 = .72 + 1.2 = 1.92 units

7 31 24.11 2.07 26.18 Step 3: FIT for month 2

8 28 27.14 2.45 29.59 FIT2 = F2 + T2
9 36 29.28 2.32 31.60 FIT2 = 12.8 + 1.92
10 - 32.48 2.68 35.16 = 14.72 units

39
 Trend Adjusted Exponential Smoothing -
Example
40 –
35 – Actual demand (At)
30 –
Product demand

25 –
20 –
15 –
10 – Forecast including trend (FITt)
5 – with a = .2 and b = .4
0 –
| | | | | | | | |
1 2 3 4 5 6 7 8 9
Time (months)

40
 Trend Adjusted Exponential Smoothing -
Exercise
 Use trend-adjusted exponential smoothing to forecast
the firm’s August income using following table and
information: Month Income (R)
Initial forecast average for February 70,000
March 68,500
February = R 65,000 April 64,800
Initial trend adjustment = 0 May 71,700
June 71,300
Smoothing constants: α = .1 July 72,800
β = .2
41
 4. Trend Projections
 A time-series forecasting method that fits a trend line
to a series of historical data points to project the line
into the future for medium to long-range forecasts.

 Least squares method is used to find linear trends:

y^ = a + bx
where y = computed value of the variable to be predicted
(dependent variable)
a = y-axis intercept
b = slope of the regression line
x = the independent variable
42
 Least Square Methods
 Equations to calculate the variables:

Least squares line: ŷ = a + bx

Slope: Intercept:

Best line!
12

10

2
Slope (b) = change in y / change in x
Intercept (a)
0
10 11 12 13 14 15 16 17 18 19 20

43
 Least Square Methods
Values of Dependent Variable (y-values)

Actual observation Deviation7

(y-value)

Deviation5 Deviation6

Deviation3
Least squares method minimizes deviations
Deviation4

Deviation1
(error) Deviation2
Trend line, y^ = a + bx

| | | | | | |
1 2 3 4 5 6 7

Time period
44
 Least Square Requirements
1. Always plot data to insure a linear relationship.

2. Do not predict time periods far beyond the given

database (20 months average data – forecast only
3 or 4 months in future).

3. Deviations around the least squares line are

assumed to be random and normally distributed,
with most observations close to the line and only a
small number farther out.
45
 Least Square Methods - Example
 The demand for electric power (MW) at a facility
over the period of 2012 to 2018 is shown in table.
Electrical Power Electrical Power
Year Year
Demand Demand
1 74 5 105
2 79 6 142
3 80 7 122
4 90

 The firm wants to forecast 2019 demand by fitting a

straight-line trend to these data.
46
 Least Square Methods - Example

Year (x) Electrical Power Demand (y) x2 xy

1 74 1 74
2
3
79
80
4
9
å 158
x 28
x =240 = =4 y=
å y

4 90 16
n
360
7 n
5 105 25 525

å x = 49 å854y = 692 = 98.86

6 142 36 852
28
7 122 x= =4 y=
Σx = 28 Σy = 692 n 2 = 140
Σx 7 n
Σxy = 3,063 7

(8) atts
4 w
10.5 mega
. 70 + r 141
6 o
= 5 1.02,
aa== yy -- bx = 98.86
98.86-10.54 ()
-10.54( 44) ==56.70
56.70
in y ear
8
=1
4
and
em
= a + bxŷ = 56.70 +10.54x
ŷThus, D

47
 Least Square Methods - Example
Trend line:
160 – y^ = 56.70 + 10.54x
150 –
Power demand (megawatts)

140 –
130 –
120 –
110 –
100 –
90 –
80 –
70 –
60 –
50 –
| | | | | | | | |
1 2 3 4 5 6 7 8 9
Year
48
 Least Square Methods - Exercise
 The table below provides the annual sales of
microwave oven for the past 9 years.
Year Sales Year Sales
1 170 6 200
2 160 7 230
3 160 8 250
4 210 9 240
5 200

Determine the forecasted sales for year 12 using

least-square method.

49
 B. Associative Forecasting
 Associative models consider several variables
(one or more independent variables) to predict the
changes in the dependent variable.

- Most common technique is linear regression

analysis.

- Extension of simple regression (multiple regression

analysis).

50
 Forecasts Used in Operations Decisions

51
52