Scribd
Upload
84 views

2b - Demand Forecasting-Some Basic Problems Solved

The document discusses three problems related to demand forecasting using exponential smoothing methods: 1. It provides the demand data for a product over the last year and uses exponential smoothing to forecast demand for the next two months, calculating errors and confidence intervals. 2. Given demand data for 8 periods, it compares exponential smoothing and simple average forecasts, finding exponential smoothing more accurate with lower errors. 3. It describes the exponential weighted moving average (EWMA) demand model and how the smoothing constant affects responsiveness vs smoothing of the forecasts.

Uploaded by

Vishvajit Kumbhar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
84 views

2b - Demand Forecasting-Some Basic Problems Solved

The document discusses three problems related to demand forecasting using exponential smoothing methods: 1. It provides the demand data for a product over the last year and uses exponential smoothing to forecast demand for the next two months, calculating errors and confidence intervals. 2. Given demand data for 8 periods, it compares exponential smoothing and simple average forecasts, finding exponential smoothing more accurate with lower errors. 3. It describes the exponential weighted moving average (EWMA) demand model and how the smoothing constant affects responsiveness vs smoothing of the forecasts.

Uploaded by

Vishvajit Kumbhar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

National Institute of Technology Calicut Department of Mechanical Engineering

Demand forecasting: Some solved problems


1. The demand manager of a company is responsible for ensuring sufficient warehouse space for a
finished item that come from the production plants. It has occasionally been necessary to rent public
warehouse space, something that the management would like to avoid. In order to estimate the space
requirements, the demand manager is evaluating exponential smoothing forecasts. The demand (in
1,000 case units) for the last fiscal year is as follows:
Month 1 2 3 4 5 6 7 8 9 10 11 12
Demand 20 18 21 25 24 27 22 30 23 20 29 22
Use a starting forecast of 20 for the month 4 to develop forecasts for month 5-12 and the first and
second month of next year. Use smoothing constant of 0.2. Calculate the mean absolute deviation
and bias. Suggest a 3-sigma interval estimate of forecast for the first and second month of next year.

Answer:

a) Average demand at a period t: 𝑋𝑡 = 𝛼𝐷𝑡 + (1 − 𝛼)𝑋𝑡−1


Forecast for period (t+1): Ft +1 = X t and for period n from t: Ft +n = X t , n = 1,2,...

Given 𝑋3=20, 𝛼 = 0.2,

Month 𝐷𝑡 𝑋𝑡 𝐹𝑡 𝐷𝑡 − 𝐹𝑡 |𝐷𝑡 − 𝐹𝑡 |
1 20
2 18
3 21 20
4 25 21 20 5 5
5 24 22 21 3 3
6 27 23 22 5 5
7 22 23 23 -1 1
8 30 24 23 7 7
9 23 24 24 -1 1
10 20 23 24 -4 4
11 29 24 23 6 6
12 22 24 24 -2 2
13 24
14 24
Xt - rounded value of Xt is provided in the above table
b) Forecast for the first and second months of next year = 24000 case units
𝐷𝑡 −𝐹𝑡 18
c) Bias: (∑𝑛𝑖=1 )= =2
𝑛 9
|𝐷𝑡 −𝐹𝑡 | 34
d) Mean absolute deviation (MAD): (∑𝑛𝑖=1 𝑛
)= 9
= 3.78
e) 3-sigma interval estimate of forecast for the first and second months of next year
Standard deviation (S) = 1.25  MAD= 1.25  3.78 = 4.72
1
Demand forecasting: Some solved problems Jan 2020
National Institute of Technology Calicut Department of Mechanical Engineering
(𝐹13 − 3 × 𝑆, 𝐹13 + 3 × 𝑆)

= (24 − 3 × 4.72, 24 + 3 × 4.72)

= (9.83, 38.16)

=(9830, 38160) case units

2. Time series data showing the demand of a product for 8 periods is given below.

Period 1 2 3 4 5 6 7 8
Demand 65 56 52 63 45 58 69 64

(a) If the above data is from a demand process that has an average demand and random error, write
the equation of the demand process.
(b) What is the estimate for average component in the forecasting equation, if the eight-period
arithmetic average is used as an estimate for average?
(c) If exponential smoothing average method with  = 0.3 and initial average = 59 is used, what is
the average at each period?
(d) What is the weight for the demand in week 6 in the forecast of week 8 if exponential smoothing
average method of forecast is used?
(e) Among the arithmetic average and exponential smoothing average methods which one you
prefer as the forecasting method for the given time series, why? What is the standard error for
each method of forecast?
(f) Calculate the bias of the forecast.

Answer:

Arithmetic average method Exponential smoothing average method


Period Dt
At Ft Dt - Ft | Dt - Ft| Xt Ft Dt - Ft | Dt - Ft|
1 65 65 60.80 59.00 6.00 6.00
2 56 60.50 65.00 -9.00 9.00 59.36 60.80 -4.80 4.80
3 52 57.67 60.50 -8.50 8.50 57.15 59.36 -7.36 7.36
4 63 59.00 57.67 5.33 5.33 58.91 57.15 5.85 5.85
5 45 56.20 59.00 -14.00 14.00 54.73 58.91 -13.91 13.91
6 58 56.50 56.20 1.80 1.80 55.71 54.73 3.27 3.27
7 69 58.29 56.50 12.50 12.50 59.70 55.71 13.29 13.29
8 64 59.00 58.29 5.71 5.71 60.99 59.70 4.30 4.30
Bias -0.88 0.83
MAD 8.12 7.35
Standard
10.15 9.18
error

Note: The notation for arithmetic average is At

2
Demand forecasting: Some solved problems Jan 2020
National Institute of Technology Calicut Department of Mechanical Engineering

a) 𝐷𝑡 = 𝑎𝑡 + 𝜀𝑡 ; where 𝑎𝑡 - Average demand, 𝜀𝑡 - Random error


b) a8 = 59
c) Exponential smoothing average
Period 1 2 3 4 5 6 7 8
Average 60.80 59.36 57.15 58.91 54.73 55.71 59.70 60.99

d) Weight for the demand in week 6 in the forecast of week 8


 (1 −  ) = 0.3  (1 − 0.3) = 0.21 (Please note that the question is to calculate the weight for
demand in the forecast of week 8, not in the average of week 8)
e)
Arithmetic Exponential smoothing
average method average method
MAD 8.12 7.35
Standard error (1.25 ×MAD) 10.15 9.18

As the standard error for the exponential smoothing average method is lower than the
standard error for the arithmetic average method, exponential smoothing average method is
preferred for forecasting.
f) Bias
Arithmetic Exponential smoothing
average method average method
Bias -0.88 0.83

3. Write an equation of the demand process suitable for simple exponentially weighted moving average
(EWMA). In this demand process, which is the factor estimated using EWMA? How are the
smoothing and responsiveness of the EWMA?
Answer:

a) 𝐷𝑡 = 𝑎𝑡 + 𝜀𝑡 ; where 𝑎𝑡 - Average demand, 𝜀𝑡 - Random error


b) a t is estimated using EWMA.
c) Smoothing and responsiveness depend on α. If α has small value better smoothing is
achieved then responsiveness is poor. For a high value of α, there will be better
responsiveness but poor smoothing.

………………………..

3
Demand forecasting: Some solved problems Jan 2020

You might also like