Chapter-9

Forecasting and Demand Planning

 Forecasting

Forecasting is the process of projecting the values of one or more

variables into the future.
• Forecasting is a key component in many types of integrated
operating systems, such as
− supply chain management systems
− customer relationship management systems
− revenue management systems.
• Poor forecasting can result in poor inventory and staffing
decisions.
• Many firms integrate forecasting with value chain and capacity
management systems to make better operational decisions.
 CPFR

CPFR Concept aims to enhance supply chain integration by supporting and assisting joint practices to reduce
extra inventory and thus quality deterioration
Need for Forecasts in a Value Chain
 Forecast Planning Horizon

Forecasts of future demand are needed at all levels of organizational

decision-making.
The planning horizon is the length of time on which a forecast is based.
• Long-range forecasts: 1 to 10 years.
• Intermediate-range forecasts: 3 to 12 months.
• Short-range forecasts: up to 3 months.
The time bucket is the unit of measure for the time period used in a
forecast.
• A time bucket can vary from one year for a long-range forecast
to an hour or less for a short-range forecast.
Linear and Nonlinear Trend Patterns
 Time Series

A time series is a set of observations measured at successive

points in time or over successive periods of time.
• A time series provides the data for understanding how the
variable to be forecast has changed historically.
• A time series pattern can be explained by five
characteristics:
− trend
− seasonal
− cyclical
− random variation
− irregular (onetime) variation.
Examples
 Judgment Forecasting

Only judgmental forecasting is possible when no historical data are

available:
• sales promotions, competitive strategies,
economic/environmental disturbances, new product
introductions, large one-time orders, etc.
Judgmental forecasting relies upon the opinions and expertise of
people in developing forecasts.
The two main approaches to judgmental forecasting are:
• Grassroots forecasting consists of asking those close to the end
consumer, such as salespeople, about the customers’ purchasing
plans.
• The Delphi method gathers individual judgments and opinions of
key personnel based on their experience and knowledge to create
a forecast.
 EXHIBIT 18.3 A GUIDE TO SELECTING
AN APPROPRIATE FORECASTING
METHOD
Amount of Forecast
Forecasting Method Data Pattern
Historical Data Horizon

6 to 12 months;
Stationary only (i.e., no
Simple moving average weekly data are Short
trend or seasonality)
often used
Weighted moving
5 to 10 observations
average and simple Stationary only Short
needed to start
exponential smoothing

Exponential smoothing with 5 to 10 observations

Stationary and trend Short
trend needed to start

Stationary, trend, and Short to

Linear regression 10 to 20 observations
seasonality medium
2 to 3 observations Stationary, trend, Short to
Trend and seasonal models
per season and seasonality medium
SIMPLE MOVING AVERAGE

A t −1 + At −2 + At −2 + ...+ At −n
Ft =
n

t- ,
SIMPLE MOVING AVERAGE –
EXAMPLE

Week Demand 3 Week 9 Week

1 800

2 1,400

3 1,000

4 1,500 1,067

5 1,500 1,300

6 1,300 1,333

7 1,800 1,433

8 1,700 1,533

9 1,300 1,600

10 1,700 1,600 1,367

SIMPLE MOVING AVERAGE –
EXAMPLE

Week Demand 3 Week 9 Week

11 1,700 1,567 1,467

12 1,500 1,567 1,500

13 2,300 1,633 1,556

14 2,300 1,833 1,644

15 2,000 2,033 1,733

16 1,700 2,200 1,811

17 1,800 2,000 1,800

18 2,200 1,833 1,811

19 1,900 1,900 1,911

20 2,400 1,967 1,933

SIMPLE MOVING AVERAGE –
EXAMPLE

Week Demand 3 Week 9 Week

21 2,400 2,167 2,011

22 2,600 2,233 2,111

23 2,000 2,467 2,144

24 2,500 2,333 2,111

25 2,600 2,367 2,167

26 2,200 2,367 2,267

27 2,200 2,433 2,311

28 2,500 2,333 2,311

29 2,400 2,300 2,378

30 2,100 2,367 2,378

SIMPLE MOVING AVERAGE –
EXAMPLE
 WEIGHTED MOVING AVERAGE

◾The simple moving average formula implies equal weighting

for all periods
◾A weighted moving average allows unequal weighting of prior
time periods
▪ The sum of the weights must be equal to one
▪ Often, more recent periods are given higher weights than periods
farther in the past
 WEIGHTED MOVING AVERAGE

Ft = w1 At −1 + W2 At −2 + ...+ w n At −n
w 1 = Weight to be given to the actual occurrence for period t -1
w 2 = Weight to be given to the actual occurrence for period t - 2
w n = Weight to be given to the actual occurrence for period t - n
n = Total number of prior periods in the forecast
 EXAMPLE

◾A department store may find that in a four month period the

best forecast is derived by using 40 percent of the actual
sales for the most recent month , 30 percent of two months
ago , 20 percent of three months ago and 10 percent of four
months ago. If the actual sales experience was as below

Month 1 Month 2 Month 3 Month 4

100 90 105 95
 Single Exponential Smoothing

Single Exponential Smoothing (SES) uses a weighted average of past

time-series values to forecast the value of the time series in the next
period.
The basic exponential smoothing model is:
𝐹𝑡+1 = 𝛼𝐴𝑡 + 1 − 𝛼 𝐹𝑡 = 𝐹𝑡 + 𝛼 𝐴𝑡 − 𝐹𝑡
Where 𝛼 is the smoothing constant, with 0 ≤ 𝛼 ≤ 1.
• Note that with exponential smoothing, 𝐹1 = 𝐴1 , and 𝐹2 = 𝐹1 =
𝐴1 .
• By repeated substitution for 𝐹𝑡 , it can be shown that 𝐹𝑡+1 is a
decreasingly weighted average of all past time-series data.
− Thus, exponential smoothing models “never forget” past
data.
 EXPONENTIAL SMOOTHING MODEL

◾Only three pieces of data are required:

1. Most recent forecast
2. Actual demand for the forecast period
3. Smoothing constant alpha ()
▪ Determines the level of smoothing and speed of
reaction
Ft = Ft −1 + α(At −1 − Ft −1 )
Ft = The exponentially smoothed forecast for period t
Ft-1 = The exponentially smoothed forecast made for the prior period
At-1 = The actual demand in the prior period
α = The desired response rate, or smoothing constant
EXPONENTIAL SMOOTHING EXAMPLE
(=0.20)

Week Demand Forecast

1 820 820

2 775 820
3 680 811
4 655 785
5 750 759
6 802 757
7 798 766
8 689 772
9 775 756
10 760
EXPONENTIAL SMOOTHING EXAMPLE
(=0.20)
Forecast 820 is
( ) (
F2 = F1 + α A − F = 820 + 0.2 820 − 820
1 1
)
Forecast 785 is
( ) (
F4 = F3 + α A − F = 811 + 0.2 680 − 811
3 3
)
Forecast 757 is
( ) (
F6 = F5 + α A − F = 759 + 0.2 750 − 759
5 5
)
Forecast 772 is
( ) (
F8 = F7 + α A − F = 766 + 0.2 798 − 766
7 7
)
Forecast 760 is
( ) (
F10 = F9 + α A − F = 756 + 0.2 775 − 756
9 9
)
 LINEAR REGRESSION ANALYSIS

◾Y = a + bt
▪ Y = Dependent variable computed by the equation
▪ y = The actual dependent variable data point
▪ a = Y intercept
▪ b = Slope of the line
▪ t = Time period
 FORECAST ERRORS

◾Forecast error is the dif ference between the forecast value

and what actually occurred
◾Can come from a variety of sources
◾All forecasts contain some level of error
◾Sources of error
▪ Bias: when a consistent mistake is made
▪ Random: errors that are not explained by the model being used
 Measures of Forecast Accuracy

Mean square error, or MSE*, is the average of the individual square

errors over all T periods of data in the time series.
𝑀𝑆𝐸 = σ 𝐴𝑡 − 𝐹𝑡 2 Τ𝑇
Mean Absolute Deviation, or MAD, is the average of the sum of the
absolute deviations for all the forecast errors in the time series.
𝑀𝐴𝐷 = σ 𝐴𝑡 − 𝐹𝑡 Τ𝑇
Mean Absolute Percentage Error, or MAPE, is the average of the
percentage error for each forecast value in the time series.
𝑀𝐴𝐷 = σ 𝐴𝑡 − 𝐹𝑡 Τ𝐴𝑡 × 100 Τ𝑇
 Example
Period Volume Forecast Error Squared Absolute Percentage
error deviation error
1 362 343.8 18.20 331.24 18.20 5.03%
2 385 361.6 23.40 547.56 23.4 6.08%
3 432 379.4 52.60 2766.76 52.6 12.18%
4 341 397.2 -56.20 3158.44 56.2 16.48%
5 382 415 -33.00 1089.00 33 8.64%
6 409 432.8 -23.80 566.44 23.8 5.82%
7 498 450.6 47.40 2246.76 47.4 9.52%
8 387 468.4 -81.40 6625.96 81.4 21.03%
9 473 486.2 -13.20 174.24 13.2 2.79%
10 513 504 9.00 81.00 9 1.75%