NATIONAL POLYTECHNIC SCHOOL

FACULTAD DE INGENIERÍA ELÉCTRICA Y ELECTRÓNICA

Production Engineering
Integrantes:Bryan Celi, Eric Chinchín, Mateo Vásquez, Joseline Yanascual
Group: Gr1
Fecha:2019-10-28
Forecasts

What is a qualitative forecasting model and when is its use appropriate?

It is that forecast in which factors such as intuition, emotions, the

personal experiences and value system of the decision-maker to arrive at the forecast.
Its use is appropriate when we do not have statistical data but we do have the
experience.

4. Briefly describe the steps to develop a forecasting system.

Determine the use of the forecast

2. Select the aspects that need to be forecasted

3. Determine the forecast horizon

4. Select the forecasting models

5. Gather the necessary data to prepare the forecast

6. Obtain the forecast

7. Validate and implement the results

5. Un administrador escéptico pregunta para qué sirve un pronóstico

in the medium term. Suggest three uses or purposes to the manager

possible.
It allows for making administrative decisions regarding planning and products, plants, and processes.

It is used to create operation plans.

6. Explain what forecasting techniques, such as moving averages, moving averages

weighted and exponential smoothing are not suitable for the data series that
they present a trend.

Weighted moving averages: in this forecasting technique, when a trend is present

the pattern, what is done is to use weights because what is needed is to give greater
importance of recent values. But it should be considered that the choice of weighting is
arbitrary, for this reason certain experience is required.
Exponential smoothing: the model is applicable to data that do not have a trend.
predictable increasing or decreasing, this due to providing a moving average with a
exponential weight of all previously observed heats giving a larger weight to
the closest observation.

9. Briefly describe the Delphi method. How could someone use it?

of the bosses for whom I have worked?

There are three types of participants in the Delphi method: decision makers, staff, and the
interviewees. The first ones are usually a group of 5 to 10 experts who will be developing the
real forecast. The staff assists decision-makers by preparing, distributing, collecting and
summarize the series of questionnaires and the results of the surveys. The interviewees form a
a group of people, often located in different places, whose judgments are valued. This group
provides information to decision-makers before making the forecast.

It can be used for long-term forecasts, avoids transportation costs.

10. What is the main difference between a time series model and a causal model?

A time series model predicts under the assumption that the future is a function of
past, while the associative or causal model incorporates the variables or factors that
they can influence the amount to be forecasted.

11. Define what a time series is

It is based on a sequence of point data separated at equal intervals. This method

it implies that future values depend only on past values, ignoring other variables,
regardless of its validity

12. What effect does the value of the smoothing constant have on the weighting given to the values?
recent

When choosing the values of the smoothing constant, the goal is to obtain the forecast.
more precise based on the most recent data

14. Which forecasting technique gives more importance to recent values? How is this achieved?

The weighted average technique gives a weighting to the observations, giving more weight to the...
importance to recent data, by giving greater weight to it.

15. Explain in your own words what an adaptable forecast is.

Signals are monitored, which when they have a variation outside of the
previously established limits, an automatic adjustment is made by computer
control signal.

18. What is the difference between a dependent variable and an independent variable?
Dependent variables: they are characteristics of reality that are determined or
that depend on the values assumed by other phenomena or independent variables.

Independent variables: The changes in the values of this type of variables

determines changes in the values of another (dependent variable).

19. Mention some examples of industries affected by seasonality. Why these?

Would businesses wish not to depend on seasonality?

One of the companies affected by seasonality is agriculture and tourism.

One of the reasons why companies would not want to depend on seasonality is because
they find it difficult to forecast their demand.

21. What happens to our ability to forecast when we forecast periods each
more distant in the future?

In long-term forecasting, several variables that can affect are being ignored.
results, such as the temporal peaks, which are unforeseen and affect the forecast, this
helps more with creation.

Problems

3. Return to problem 4.2. Develop a forecast for years 2 to 12 using smoothing.

exponential with α = 0.4 and a forecast for year 1 of 6. Graph your new forecast along with
with real data and an intuitive forecast. Based on a visual inspection, what forecast
Is it better?

= t-1 + (t-1 − t-1 )

Year Demand Forecast Dream-Forecast

1 7 6.00 1.00
2 9 6.40 2.60
3 5 7.44 2.44
4 9 6.46 2.54
5 13 7.48 5.52
6 8 9.69 1.69
7 12 9.01 2.99
8 13 10.21 2.70
9 9 11.32 2.32
10 11 10.39 0.61
11 7 10.64 3.64
12 9.18

The best forecast is for year 9 with 11.32

 Graph Title
14

12

10

0
1 2 3 4 5 6 7 8 9 10 11

Demand Forecast

6. Monthly sales at Telco Batteries, Inc. were as follows:

a) Graph the monthly sales on graph paper.

b) Forecast sales for January using each of the following:
i. Intuitive method.
ii. 3-month moving average.
iii. Weighted moving average of 6 months with .1, .1, .1, .2, .2, and .3, applying the
higher weights to the most recent months.
iv. Exponential smoothing with α= .3 and a forecast for September of 18.
v. A trend projection.
c) With the given data, what method would allow you to prepare the sales forecast for the
next month of March?
a) The graphs were obtained in Excel
 Telco Batteries Inc. Sales
25

20

15

10

b)
i. It is anticipated that there will be 23 sales in January.
ii. Moving average

Months Sales Forecast

October 20
November 21
December 23 20 + 21 + 23 = 64 / 3 = 21.3
January 21.3

iii. Weighted moving average

Months Sales Weighting Value
July 17 0.1 1.7
August 18 0.1 1.8
September 20 0.1 2
October 20 0.2 4
November 21 0.2 4.2
December 23 0.3 6.9
January =1.7+1.8+2+4+4.2+6.9=20.6
iv. Exponential smoothing

= t-1 + ( t-1 − t-1 )

Months Sales Forecast
September 20 18
October 20 18.6
November 21 19.2
December 23 19.61
 January 20.627
v. Trend

̂= +

∑ − ̅̅
= 2−
∑ ̅2

x̅ = 6.5 ; ̅ = 18.167∑ = 1474∑ 2650

= 1.8275
= ̅ − bx̅ = 6.288
̂( ) = 30.05
c) It is recommended to use the exponential smoothing method, as it is closer to a
value that is within the sales

9. Resume the solved problem 4.1 on page 140. Use a 3-year moving average to
forecast the sales of Volkswagen Beetle in Nevada during 2008. What is the MAD?

Problem solved 4.1

The sales of the popular Volkswagen (VW) Beetle have grown steadily.
during the last 5 years in the establishments of Nevada (see the table below).
The sales manager forecasted in 2002 that sales for 2003 would be 410 VW.
exponential smoothing with a weighting α = .30 to develop the
forecasts from 2004 to 2008.

Year Sales Forecast Forecast error

2003 450 410

2004 495

2005 518

2 75.33
2006 563 487
3
1 58.67
2007 584 525
3

2008 ? 555
 450 + 495 + 518 2
ó 2006 = = 487
3 3
495 + 518 + 563 1
ó 2007 = = 525
3 3

518 + 563 + 584

ó 2008 = = 555
3
75.33 + 58.67
= 67
2

12. In problems 4.10 and 4.11, three forecasts of the enrollments were developed.
seminar on: 3-year moving average, weighted moving average, and smoothing
exponential. With MAD as a criterion, which of the three forecasting methods is better? Explain.
your answer.

Moving average Smoothing

Moving average
Año Inscripciones(miles) weighted exponential

Forecast Error Forecast Error Forecast Error

1 4 5,00 1.00

2 6 4,70 1.30

3 4 5.09 1.09

4 5 4.67 0.33 4.50 0.50 4,76 0.24

5 10 5,00 5,00 5.00 5.00 4.83 5.17

6 8 6.33 1.67 7.25 0.75 6.38 1.62

7 7 7.67 0.67 7.75 0.75 6.87 0.13

8 9 8.33 0.67 8.00 1.00 6.91 2.09

9 12 8.00 4.00 8.25 3.75 7.54 4.46

10 14 9.33 4.67 10.00 4.00 8,87 5,13

11 15 11.67 3.33 12.25 2.75 10.41 4,59

11 13.67 14.00 11.79

MAD 2.54 2.31 2.44

According to the MAD values, the best technique is the moving average, which indicates that the
The forecasts obtained do not deviate from the actual values.

15. Take up the solved problem 4.1 from page 140. Use a 3-year moving average to
forecast the sales of Volkswagen Beetle in Nevada during 2008. What is the MAD?

∑
ó : −−−− =3

Year Sales Average

2003 450
2004 495
2005 518 487.67
2006 563 525.33
2007 584 555
2008 ?

∑| − | ( 518− 487.67 +) 563

( − 525.33 +(584−
) 555)
= = = 32.33
3

18. Return to problem 4.1 and to problems 4.15 and 4.16. With MAD as the criterion.
Would you use exponential smoothing with a smoothing constant of .3 as shown in the
problem solved 4.1, a 3-year moving average, or I would use the trend to predict the
sales of Volkswagen Beetle? Explain your answer

∑| |
=

Method MAD
Exponential smoothing 74.55
Moving average 32.33
Trend 41.2

In the table, it can be observed that the moving average method has a smaller error, therefore,
moving average would be used to forecast sales.

24. The operations manager of a musical instrument distributor believes that the
demand for drums may be related to the number of television appearances of
popular rock group Green Shades during the past month. The manager has gathered the
data shown in the following table:

a) Create a graph with this data to see if a linear equation could describe the relationship
between the group's appearances on TV and the sale of drums. b) Use the regression method of
least squares to obtain a forecasting equation. c) What would be your estimate of the
Sales of drums if Green Shades had appeared 9 times on TV last month?

Demand for drums vs

Television appearances
12
10
8
6
4
2
0
0 2 4 6 8 10

b.

Appearances in
Demand of
Green TV
bombs x squared
Shades xy

3 3 9 9

4 6 24 36

7 7 49 49

6 5 30 25

8 10 80 100

5 8 40 64

x̅ = 5.5 ̅ = 6.5 ∑ = 232 ∑ = 199

232 − 6(5,5)(6,5)
= =1
( )2
199− 6 5.5

a = 6.5- 1 5.5( =1)

 = 1+
c.

= 10
The sales estimate would be 10 drums

The table presents the data on the number of transistors (in millions) manufactured in a
Japan's plant over the last 5 years:

a) Predict the number of transistors that will be manufactured next year using regression.
linear.
b) Calculate the mean squared error (MSE) when using linear regression.
c) Calculate the mean absolute percentage error (MAPE).

a) Linear regression

̂= +
∑ − ̅̅
= 2−
∑ ̅2

x̅ = 3 ; ̅ = 180;∑ = 2880∑ 2= 55

= 18
= ̅− ̅ = 126
̂ (6 )= 234
b) MSE
2
∑| |
= = 32

c) MAPE

| − |
100 ∑
= = 2.65
36. The accountants of the firm Doke and Reed believe that some executives were including
in their accounts, expense vouchers of unusually high amounts when they returned from their trips.
business. First, they took a sample of 200 vouchers issued during last year.
They then developed the following multiple regression equation to relate the cost.
expected from the trip with the number of travel days (x1) and the distance traveled (x2) in miles:

= + . + .
The calculated correlation coefficient was 0.68.

If Bill Tomlinson returns from a 300-mile trip that took him 5 days away from the city, what
Is it the expected amount that should be requested for expenses? b) Tomlinson submitted a request for
Refund of 685 dollars. What should the accountant do? c) Should other variables be included?
Which ones? Why?

a.

y = 90 + 48.505 ( +) 0.40 300

( = $452.50
)
b.

The accountant must review the additional expenses that Tomlinson considered because there is a
difference between what was requested and the actual value obtained in the first item

An additional percentage of the total must be included in case of unforeseen events.

39. Dr. Susan Sweeney, a psychologist based in Providence, specializes in treatment.

of patients with agoraphobia (i.e., fear of leaving home). The following table indicates how many
patients attended annually over the last 10 years. It also indicates the index of
robots registered in Providence for the same year.

Using trend analysis, forecast the number of patients that Dr. Sweeney
It will serve in years 11 and 12. How well does the model fit the data?
= +

Year (x) Patients(y) 2 xy

1 36 1 36
2 33 4 66
3 40 9 120
4 41 16 164
5 40 25 200
6 55 36 330
7 60 49 420
8 54 64 432
9 58 81 522
∑ x = 45 ∑ y = 417 ∑ 2= 285 ∑ = 2290
 ∑ 45 ∑
̅= = 9
=5 ̅= = 417
9
= 46.33
∑ − ( )(
̅ ̅ 2290− 9 546.33 )
= = = 3.419
∑ 2− ̅ 2 285− 925( )
= ̅ − bx̅ = 46.33− 3.419 5 =( 29.235
)

= 29.235 + 3.419x
( 11=) 29.235 +113.419( = 66.84
)
( 12=) 29.235 +123.419( =70.26
)

42. Des Moines Power and Light has collected the data on the demand for electric power in its
western subregion only for the last 2 years. The information is shown in the following table

To plan an expansion and agree on the energy loan with other service companies
public during peak periods, Des Moines needs to be able to forecast the
demand for each month for the next year. However, the standard forecasting models
analyzed in this chapter do not fit the data observed over the 2 years.

What are the weaknesses of standard forecasting techniques when applied to this series?
data?

El análisis se basa en intuición y análisis de datos previos, por lo tanto se puede

to assert that the technique is inaccurate since it does not present an argument
strong mathematician

Since the known models are not really suitable for this case, please make a proposal.
to approach the forecast. Although there is no perfect solution for managing data from

this type (in other words, there are no answers that are 100% correct or incorrect),
justify your model.

It is suggested to use the weighted method. With this method, what is done is
determine an average by adding the existing values in the inventory with
values of the new purchases, to then divide it by the number of units
existing in the inventory including both those initially existing and those of
the new purchase
Forecast the demand for each month of next year with the model you proposed.
45. The Dubuque East School District is trying to forecast its teacher needs.
kindergarten for the next 5 years. The district has birth data and
registrations from the last 10 years. The registrations occur 5 years after the births.
Forecast the enrollments for the next 2 years, according to the data from the last
10 years:

Linear regression is used for birth forecasting.

̂= +
∑ − ̅̅
= 2−
∑ ̅2

x̅ = 5.5 ; ̅ = 130.7∑ xy = 6907;∑ 2= 385

= −3.412
= ̅− ̅ = 111.9333
̂ (11=)74.401
̂ (12=)70.989
For the forecast of enrollments, multiple regression is used, where data was inputted.
to Excel to obtain the parameters of the equation:

̂= + 1 1+ 2 2

Where x1 is for the year and x2 is for births.

= 147.8945

1= -1.08116

2= −0.0149
( 11,74.401=) 134.89
( 1270.989 )= 133.8628
48. Sundar Balakrishnan, general manager of Precision Engineering Corporation (PEC), believes that
the engineering services that road construction companies hire your
the company is directly related to the business volume of road construction
that hires companies in its geographical area. It wonders if it really is this way, and if
It is, would this information help plan your operations better if you forecast the amount
of the engineering services required by construction companies in each quarter of
year? The following table presents the sales data of your services and the total amounts
of the construction contracts for roads from the last 8 quarters:

a) With this data, develop a regression equation to predict the level of demand for
the PEC services.

Trimestre Ventas servicios de PEC (en miles) Contratos firmados (en miles de dólares)
1 8 153
2 10 172
3 15 197
4 9 178
5 12 185
6 13 199
7 12 205
8 16 226

Coefficients

Interception -21.5848283

Variable X 1 -0.86852477

Variable X 2 0.19732377

=− . − . + .