review of combining forecasts
approaches
Aline Castello Branco Mancuso
Federal University of Rio Grande do Sul (UFRGS) -
Brasil
E-mail: aline.mancuso@gmail.com
Liane Werner
Federal University of Rio Grande do Sul (UFRGS) -
Brasil
E-mail: liane@producao.ufrgs.br
Submission: 09/03/2013
Accept: 11/03/2013
ABSTRACT
The first review of the literature
on the subject combination of forecasts was made in the twentieth century by
Robert Clemen. After more than twenty years, several other papers have been published
with new theories and applications, but no other similar review was performed.
Faced with this placement, this paper aimed to review the literature on the
approaches of combining forecast after the survey conducted by Clemen (1989),
covering the various areas of knowledge. Thus, this paper presents the
classification and analysis of 174 articles collected on the subject,
describing their main characteristics. As main contributions, this paper
offers: a summary of current literature on the topic; a classification of
articles according to the approaches; a subdivision of items within each
approach; analysis of classification and identification of the most common
methods, new methods, and future research.
Keywords: combining
forecasts, review of the literature, forecasting.
1. INTRODUCTION
The
market is constantly changing and evolving, with local actions depending on
global phenomena in addition to increasingly fierce competition. Faced with
this scenario, formulation and strategic planning become vital to
organizational survival and growth, both in theoretical and practical areas.
Attentive to the new world order, as well as to business needs, search for
optimization of the quantity of products to be produced with as minimal waste
as possible has become the main topic of many studies.
Industries
need accurate demand forecasts, since any significant deviation of the actual
demand can cause several types of negative impacts, especially to the economic
performance of the company. Besides the storage cost and risk of product
obsolescence, some business sectors, such as food, for example, need to
consider the short duration of perishable products, among many other variables
that influence the volume and quality of production. However, it is essential
to find a balance to avoid overproduction being wasted or else disadvantage
caused by low productivity (CASAGRANDE & HOSS, 2010).
Due
to this increasing need of greater control over the storage level, a mechanism
has been designed to connect budget and actual demand. Demand forecasting is a
process for making statements about events whose results have not yet been
observed; after all, the future is unknown, but not unpredictable.
Consequently, forecasting became an indispensable activity in planning, setting
strategy and making business decisions.
Initially,
the prediction process was based on a single technique, among the various
options available. However, in many situations, single results may not be
sufficient for decision making, due to the market complexity. Aiming at
increasingly reliable forecasts along with the lowest error prospect, a
procedure to add and adjust various forecasting techniques was developed,
namely the integration of forecasts (WEBBY & O´CONNOR, 1996). Among
different integrated structures, this study is limited to a method known as
combining forecasts. It is a method that uses some objective or subjective
mechanism to compose forecasts to obtain a final forecast (combined). According
to Clemen (1989), among many ways developed to perform combining forecasts, so
far, the results have been unanimous: combining forecasts lead to increased
accuracy.
In
the twentieth century, Clemen (1989) conducted the first review of the
literature on the subject combining forecasts, listing 209 broad aspect publications.
After more than twenty years, several other studies have been published with
new theories and applications. However, none of those works compares to the one
published by that author. Faced with this situation, the purpose of this paper
is to review the literature on the approaches of combining forecasts, following
the survey conducted in 1989 by Robert Clemen, state of the art in the area up
to that date.
This
article is divided into five sections, the first being this introduction.
Section two presents the methodological procedures for the development of this
article. Section three presents a theoretical combination of forecasts. The
classification and analysis of mapped items are presented in section four.
Finally, section five presents the main conclusions of this study.
2. METHODOLOGICAL PROCEDURES
Revision
of specific literature review aims to survey and analyze material published on
a particular topic, identifying the gaps and the main theoretical or
methodological obstacles. This research provides a mapping of what has been
written and by whom (SILVA & MENEZES, 2001). By synthesizing primary similar studies, the
secondary ones serve as support for the targeted search, summarizing the volume
of literature.
Hence,
this section provides detailed steps of the literature review procedure. The
project design ensures research reproducibility, besides allowing comments,
suggestions and criticisms to the method used.
The
realization of the review initiates with the formulation and definition of the
problem, enclosing the material to be analyzed. The completion of the review
starts with the formulation and problem definition, which enclose the material
to be analyzed. In this work, we want to obtain information about the current
state of the literature of combining forecasts, as well as its use. For that
reason, the theoretical framework should clarify the issue, providing a better
definition of the subject studied.
The
search for studies was conducted online during the year 2012. Through the
exploration of the ten areas of knowledge recorded in Journal Portal Capes, 83
databases were filtered to be searched in full-text journals. Consulting
databases from such periodicals, the articles selected were the ones which
mentioned the words "combining" and "forecast" or
"combine" and "forecasting” in their title or keywords. Since
the main objective of this work is to examine studies on combination of
forecasts, continuing the review by Clemen (1989), only studies published from
1989 on were compiled. The search resulted in 256 articles, except for the work
already covered by Clemen (1989), and the studies on the combination of models
and inference combinations (combination of intervals and densities), considered
outside the scope of the search. The final number of identified articles was
reduced to 174. The total figure resulting from the search, as the strategies
used for each database, can be seen in Table 1.
Table 1.
Articles identified in Databases.
Databases |
Publisher |
URL |
Articles |
ACM Digital Library |
Association for Computing Machinery |
23 |
|
IEEE Xplore |
IEEE |
07 |
|
Cambridge Journals Online |
Cambridge University Press |
02 |
|
OECD iLibrary: Periodicals |
Organization for Economic Co-operation and Development |
http://www.oecd-ilibrary.org |
01 |
American Physical Society |
American Physical Society |
01 |
|
SciElo |
Bireme |
02 |
|
AIP Scitation |
The American Institute of Physics |
01 |
|
Central Online and Open Access Library |
Copernicus Systems + Technology GmbH |
01 |
|
Academic Search Premier |
EBSCO |
03 |
|
Social Sciences Full-Text |
HW Wilson |
02 |
|
High Wire: Free Online Full Text Articles |
High Wire Press |
03 |
|
Maney Publishing |
|
52 |
|
Wiley Online Library |
Wiley Inter Science |
33 |
|
Cell Press Collection |
Cell Press (Elsevier) |
44 |
After defining the research question
(stage I), and mapping available literature (stage II), approaches were
identified (stage III), following the classification of articles in approaches
(stage IV) and the analysis of the classification of approaches (stage V).
However, before the mapping, a theoretical framework on the subject must be
presented.
3. COMBINATION OF FORECASTS
In
the production of goods, demand forecasting can be understood as a function
related to predicting the consumption of products, so that they can be
appropriately manufactured to meet demand. However, this function can be
described by several methods, ranging from informal judgment, intuition and
expert opinion, through macro-economic factors, and even forecasting techniques
based on historic data.
Most
of the methods employed analyze information using only a single forecasting
technique; as a consequence, some information from other techniques end up
disregarded (WERNER & RIBEIRO, 2006).
As previously mentioned, the complexity of the market demands all
available information for forecasting, and a single technique cannot make
efficient use of such great deal of information. Since Bates and Granger
(1969), the fact that forecasting techniques can become more accurate when
performing combination has been studied. Regardless of how the combination is
obtained, its result is intended to cause an increase in the accuracy of the
individual estimates. This happens because individual forecasting techniques
are based on different approaches, and can therefore capture distinctive
characteristics of the series as well as allow the combination to benefit from
such characteristics (ARMSTRONG, 2001).
The
idea of combining forecasts is simple, as described in Figure 1. Based on a
set of information, the forecasting models are generated based on different
techniques (technique 1, technique 2, …, technique n), providing n forecasts.
Such forecasts are then combined, generating a single final forecast. Armstrong
(2001) discusses the number of techniques to be considered in combination,
concluding that, with respect to efficiency, five would be suitable. The author
bases his suggestion on the exponential behavior of the combination gains. The combination of five forecasts reduces the
amount of errors, but when more than five techniques are combined, gains get
smaller and smaller at each addition.
However,
the question is: How should such techniques be combined? Two combination
approaches are defined in the literature: one involving an objective approach
and another involving a subjective one. The objective approach represents
methods which use a mathematical function, so that results can be repeated. The
subjective approach includes intuitive efforts to combine forecasts, by means
of knowledge and opinion.
Figure 1. Combination of Forecasts Source: adapted from Webby and O’Connor (1996, p.
100). |
Objective
methods of combination originated with Bates and Granger (1969), considered the
forerunners of this subject. They proposed the method to combine the forecasts
through a linear combination of two non-biased objective forecasts (or properly
corrected) considering a weight k for the first, and (1 – k) for the second, as depicted in
Equation (1).
(1)
Where:
is the value of the combination, the value of forecast one, the value of forecast two and is the factor that minimizes the error
variance of the combined forecast.
Subsequently,
other authors adhered to the method and studies advanced in the area. The
combination of forecasts was extended from two to “n”, and combined techniques
began to be interpreted as a structured form of regression (NEWBOLD &
GRANGER, 1974). Since then, several authors have suggested new considerations
and more sophisticated methods have been compared. However, the arithmetic mean
is still one of the most commonly used methods (MENEZES et al., 2000).
An
example of arithmetic mean performance compared to other methods of combination
can be seen in Marques (2005). The author considers the simple average, the
average weighted by the inverse of the mean square error (equivalent to the
minimum variance method), the optimization with weight restriction and without
constant (which is the estimated weights by the method of least squares, with
weight restriction and without constant) and the optimization without weight
restriction, with constant.
Various
forms of combination forecasts have been developed since the publication of the
article by Bates and Granger (1969), extending from the simple arithmetic mean
to more sophisticated methods such as neural networks to nonlinear combinations
or studies using Bayesian analysis for the combining forecast, which in general
weighs each forecast based on the expected value. Chan et al. (1999) listed
some classic works in this approach. However, there is no consensus in the
literature that a sophisticated combination method might be superior to simpler
ones, as the average of individual forecasts. Like Clemen (1989), Werner (2005)
emphasizes the combination via simple average which, despite lacking optimal
weights, can provide better results compared to more sophisticated methods.
Nonetheless, Martins (2011), for example, discusses the superior performance of
the combination via minimum variance.
Moreover,
to establish the accuracy of a forecast in the objective approach, it is
essential to estimate error. Paliwal and Kumar (2009) observed the use of MAPE
(Mean Absolute Error Perceptual), MSE (Mean Square Error) and MAE (Mean
Absolute Error) as the main measures to evaluate model performance in several
studies. However, variations as the Root Mean Square Error (RMSE), among
others, are also commonly applied.
The
subjective approach of combination is still considered unexplored, given that
intuition can hardly be repeated (WERNER & RIBEIRO, 2006). It is usually
used with scarce data, while launching a new product, for example. Using only
intuition and acquired knowledge, as the consensus practices of a group, the
Delphi method and the selection of the best experts are distinguished for
combining forecasts subjectively. For Armstrong (2001), in most situations, the
first step should be the opinion of experts.
The
combination of models (objective techniques) with human judgment (subjective
techniques) follows the same principle of combination mentioned above, and
illustrated by Figure 1. However, this method is best seen in. Based on historical
data, the model is generated at the same time as human judgment is executed,
adding contextual information to yield two forecasts (objective and
subjective). These forecasts are combined and, based on contextual information,
one single final forecast is generated (WEBBY & O’CONNOR, 1996; WERNER,
2005).
Werner
(2005) discusses some publications based on subjective combining, concluding
that combining forecasts are influenced by the individual characteristics of
predictors, as well as by the aspects of the forecasting context.
Figure 2. Subjective Combining Method Source: Webby and O’Connor (1996, p. 100). |
There
are many studies on combining forecasts in the literature, as represented in
the mapping. Moreover, this popularity of combination is due to the fact that,
instead of attempting to choose the best technique, the problem is formulated
by asking which techniques could improve accuracy (WERNER, 2005; WERNER &
RIBEIRO, 2006). Armstrong (2001) recommends combining forecasts when there is
no assurance about the situation and/or technique precision, so as to avoid
significant mistakes.
4. MAPPED ARTICLES
In an
analysis preceding the identification of the approaches, obtained data were
described according to year of publication and journal. Table 2 presents the
number of articles per journal, being the category "others" the group
of journals that presented only one publication during the period surveyed.
Table
2 shows that the periodicals International Journal of Forecasting and Journal
of Forecasting concentrated approximately 37% of the publications relating to
combining forecasts in that period. Still, yearly publications are better
analyzed in. Additionally, Figure 3 shows a positive trend over the years,
especially 2007 and 2008, both with 14 publications. The culmination of the
work in this area was year 2011, with 23 publications.
Table 2. Number
of articles per journal and year of publication.
|
1989 |
1990 |
1991 |
1992 |
1993 |
1994 |
1995 |
1996 |
1997 |
1998 |
1999 |
2000 |
2001 |
2002 |
2003 |
2004 |
2005 |
2006 |
2007 |
2008 |
2009 |
2010 |
2011 |
2012 |
Total (%) |
Applied Economics |
|
|
|
|
|
|
|
|
1 |
|
1 |
1 |
|
3 (1,72) |
|||||||||||
Applied Economics Letters |
|
|
|
|
|
|
|
|
1 |
1 |
|
2 (1,15) |
|||||||||||||
Computers and Industrial Engineering |
|
|
|
|
|
|
|
|
1 |
|
1 |
|
2 (1,15) |
||||||||||||
Conference |
|
|
|
|
|
|
|
|
|
1 |
1 |
1 |
4 |
1 |
1 |
2 |
|
11 (6,32) |
|||||||
Decision Sciences |
1 |
|
|
|
2 (1,15) |
||||||||||||||||||||
European Journal of Operational Research |
|
|
1 |
|
1 |
1 |
1 |
1 |
|
6 (3,45) |
|||||||||||||||
Expert Systems with Applications |
|
|
1 |
2 |
1 |
4 (2,30) |
|||||||||||||||||||
Hydrological Processes |
|
|
1 |
1 |
2 (1,15) |
||||||||||||||||||||
International Journal of Forecasting |
6 |
1 |
|
|
1 |
|
1 |
3 |
1 |
3 |
2 |
1 |
2 |
1 |
2 |
1 |
1 |
2 |
3 |
1 |
7 |
1 |
45
(25,86) |
||
International J. of Production Economics |
0 |
0 |
1 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
2 |
|
2 (1,15) |
||||
J. Computational Statistics & Data
Analysis |
1 |
1 |
1 |
|
2 (1,15) |
||||||||||||||||||||
Journal Decision Analysis |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Journal Management Science |
1 |
1 |
1 |
1 |
|
4 (2,30) |
|||||||||||||||||||
Journal of Applied Statistics |
1 |
1 |
1 |
1 |
|
4 (2,30) |
|||||||||||||||||||
Journal of Forecasting |
2 |
2 |
3 |
2 |
1 |
1 |
1 |
1 |
3 |
1 |
2 |
|
19
(10,92) |
||||||||||||
Journal of Hydrology |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Journal of the Operational Research
Society |
1 |
2 |
|
3 (1,72) |
|||||||||||||||||||||
Journal of Time Series Analysis |
1 |
1 |
2 (1,15) |
||||||||||||||||||||||
Management Science |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Mathematical and Computer Modeling |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Meteorological Applications |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Oxford Bulletin of Economics and Statistics |
2 |
1 |
3 (1,72) |
||||||||||||||||||||||
Technological Forecasting and Social
Change |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Tourism Management |
1 |
1 |
|
2 (1,15) |
|||||||||||||||||||||
Others |
1 |
1 |
1 |
1 |
0 |
3 |
1 |
1 |
1 |
3 |
1 |
2 |
1 |
1 |
2 |
3 |
0 |
3 |
6 |
3 |
5 |
3 |
1 |
44 (25,29) |
|
Total (%) |
8 |
4 |
2 |
8 |
4 |
3 |
7 |
7 |
5 |
5 |
12 |
5 |
5 |
4 |
3 |
7 |
7 |
5 |
14 |
14 |
10 |
6 |
23 |
6 |
174 |
(4,60) |
(2,30) |
(1,15) |
(4,60) |
(2,30) |
(1,72) |
(4,02) |
(4,02) |
(2,87) |
(2,87) |
(6,90) |
(2,87) |
(2,87) |
(2,30) |
(1,72) |
(4,02) |
(4,02) |
(2,87) |
(8,05) |
(8,05) |
(5,75) |
(3,45) |
(13,22) |
(3,45) |
Figure 3. Temporal evolution of publications Source: prepared by the author |
Regarding
approaches, we identified four major groups: a summary or review of the
literature, comparison of methods, application and exploration.
From
a total of 174 articles compiled, 80 were considered exploratory, 73 were
applicative, 15 were selected for the comparison of methods and 6 are abstracts
or reviews. What follows is the importance to emphasize that these ratings are
suggestive. Many publications do not present an exact definition of attributes;
some articles, for example, use both application and comparison in the
discussion of combining methods. However, in such cases, the suggested
classification was based on the content of these studies: summary and review of
the literature studies that prioritize the review of the subject; comparison of
methods and works confronting methods; applications to compare or explore the
use of combination applied to a specific topic; and exploratory, which are
designated to the theoretical analysis.
During
the mapped period, four review articles and two abstracts were found. Clemen
(1989) provided an annotative bibliography of the literature so far, including
the contributions of psychology, statistics and other areas of knowledge, in
addition to highlighting some suggestions for further studies. Smith (1997)
reviewed and discussed the performance of simple models compared to more
complex ones. Yet, in the study of model precision, Menezes et al. (2000) have
also revised the guidelines for the use of the combination and analyze data on
the performance of different combining methods, aiming to provide practical
guidance based on three properties of forecast errors: variance, asymmetry and
correlation. More recently, Jorgensen (2007) examined subjective combining.
Nevertheless,
Armstrong (2006) and Wallis (2011) defined that work as a summary. While
Armstrong (2006) summarizes the progress made during the past 25 years,
regarding methods to reduce forecast error.
Wallis (2011) dedicates his review to the memory of Clive W. J. Granger
(1934 - 2009), resuming some of the themes approached in the article of Bates
and Granger (1969).
Several
authors compare their results with results of other authors in an attempt to
suggest a new method or justify the proposed one. However, this kind of
approach lists works that assign a larger space between the comparison
techniques and models. In Table 3, 15 publications classified in this area are
identified (first column) along with the methods compared (second column).
Columns three and four distinguish the focus of comparison (context), when it
is set, and the specific application area, respectively.
Despite
the comparative content of these works, some authors focus on the application
areas. Such principle of comparison and application highlights the prominent
comprehensiveness of economy. In the macroeconomic area, more specifically,
Walz and Walz (1989) confront a Bayesian method of combination with multiple
regressions. Moreover, applied to exchange rates, Tsangari (2007) analyzes
different methods of combining forecasts and proposes an alternative
methodology. Yet, Jing-Rong (2007) and Jing-Rong (2008) compares the combining
forecast model proposed with the individual forecasts and the combination of
methods. In both works the author uses forecasts of stock market volatility;
the difference resides in the Bayesian interpretation of the combining forecast
suggested by the author in 2008. Furthermore, currently, in the field of
macroeconomics, Poncela et al. (2011) compare the use of four combination
methods: most important components, dynamic factor models, partial least
squares and sliced inverse regression. In tourism, Shen et al. (2008) analyze
three combining methods: simple mean, variance-covariance (minimum variance)
and the discounted mean square forecast error method.
Table 3.
Articles mapped in attribute Comparison of Methods.
Methods Compared |
Context |
Application |
||||||||
Walz e Walz (1989) |
- multiple regression - Bayesian method |
Bayesian method |
macroeconomic |
|||||||
Lobo e Nair (1990) |
- individual forecasts |
equal weights versus unequal ones |
annual profit accounting |
|||||||
- objective methods |
||||||||||
- subjective methods |
||||||||||
Aksu e Gunter (1992) |
Four combination methods |
|
||||||||
Ming Shi et al. (1999) |
- individual forecasts - linear methods - nonlinear methods |
ANN (artificial neural
networks) |
||||||||
Taylor e Bunn (1999) |
- theoretical methods |
nonparametric for accurate measurements |
|
|||||||
- empirical methods |
||||||||||
Goodwin (2000) |
Three integration methods subjective |
|||||||||
Jing-Rong (2002) |
- individual forecasts - conventional methods |
FNN (Fuzzy Neural Network) |
|
|||||||
- nonlinear methods |
||||||||||
Jing-Rong (2007) |
- individual forecasts |
|
forecasts of stock market volatility |
|||||||
- combining forecast - combination of methods |
|
|||||||||
Tsangari (2007) |
Several |
|
exchange rates |
|||||||
Jing-Rong (2008) |
- individual forecasts |
Bayesian method |
forecasts of stock market volatility |
|||||||
- combining forecast - combination of methods |
||||||||||
Shen et
al. (2008) |
- individual forecasts - three combining
methods |
tourism |
||||||||
Jeong e Kim (2009) |
Eight combination methods |
|
||||||||
Hsiao e Wan (2011) |
Two combination methods |
scenarios prone to structural breakage |
||||||||
Poncela et al. (2011) |
Four combination methods |
macroeconomics |
||||||||
Martins e Werner (2012) |
- individual forecasts - Three combination
methods |
correlation between the errors |
industrial series |
|||||||
Besides
comparing methods, several authors compare combining with individual forecasts.
Shi et al. (1999) discuss the use of artificial neural networks (ANN) in
relation to methods of linear combination and Jing-Rong (2002) compares fuzzy
neural networks (FNN) with conventional methods. Yet, applied to the forecast
annual profit accounting, Lobo and Nair (1990) also compare combining with
individual forecast, and discuss objective in opposition to subjective methods.
Additionally, they analyze the use of equal weights versus unequal ones.
And
in the scope of exploration, Aksu and Gunter (1992) comment on the efficiency
of the simple mean, of the OLS (Ordinary Least Squares) model, ERLS - Equality
Restricted Least Squares (also Restricted Least Squares) and NRLS - Restricted
Non Negative least Squares. Taylor and Bunn (1999) compare the ability of
theoretical and empirical methods, and a nonparametric proposal for accurate
measurements. In the field of subjectivity, Goodwin (2000) compares the
accuracy of combined forecasts with two other integration methods. Jeong and
Kim (2009), from theoretical angles, contrast the two most popular combining
methods, simple average and weighted average, with six others. The authors also
develop a guideline for the choice of combining method using analytical
derivations.
In a
more contemporary context, some authors use the comparison between methods as
an exploration basis. Martins and Werner (2012) intend to identify differences
in the accuracy of forecasts obtained with and without considering the
correlation between the errors, comparing individual forecasts to the
combination by simple average and minimum variance with and without
correlations. Hsiao and Wan (2011), however, suggest two corrections for the
combination via simple average, comparing models in scenarios prone to
structural breakage.
Moreover,
this class includes an extremely well-known forecast competition: M-Competition
(MAKRIDAKIS et al., 1982) and its subsequent versions, the M2-Competition (MAKRIDAKIS
et al., 1993) and M3-Competition (MAKRIDAKIS & HIBON, 2000). The main idea
of the authors is to hold a competition with the largest possible number of
sets and models. A set of 1001 series is used in the M-Competition, with
interest models forecasting these series (24 different models). Later versions
include new series and new models in the competition. However, these three
publications of the Competition series do not take part of the group of papers
screened, as they go beyond the scope of this study. Addressing them, however,
is essential in any reference work on combining forecasts.
In
the application criterion, there are 73 papers that analyze and/or use the
combination of forecasts as a forecasting method for a particular field. Table
4 identifies such field and their references.
According
to Table 4, in this attribute there is a significant amount of work from the
field of economy, with 26 publications. Weather has the second highest rate,
with 12 publications.
As in
the previous attribute, applied works are not necessarily unique. Many authors
have used comparisons and engaged in an exploration. Among the references
identified as application stand out Xiong et al. (2001), Dreger and Schumacher
(2005) and Rapach and Strauss (2008), on a criterion of comparisons. Menezes
and Bunn (1993), Volkov and Gladkov (1995), Terregrossa (2005), and Moreno and
Lopez (2007) and Moreno and Lopez (2011) and Christodoulos et al. (2011) also
have works distinguished by the level of exploratory application. Lubecke et al.
(1998) are prominent as well in the exploration of neural networks.
Table 4.
Articles mapped in attribute Application.
Field |
References |
Context |
||
Economy and Finance |
Spiro
(1989) |
Macroeconomic / Canada |
||
Lobo
(1991) |
Net business |
|||
Collopy
e Armstrong (1992) |
Economic and
demographic Series |
|||
Klein
e Park (1993) |
USA |
|||
Menezes
e Bunn (1993) |
Annual inflation / UK |
|||
Macdonald
e Marsh (1994) |
Exchange Rates |
|||
Volkov
e Gladkov (1995) |
Exchange Rates |
|||
Donaldson
e Kamstra (1996) |
Stock market
volatility |
|||
Shen
(1996) |
Macroeconomic /
Taiwan |
|||
Lubecke
et al. (1998) |
Exchange rate
volatility |
|||
Hu
e Tsoukalas (1999) |
European Monetary
System (EMS) |
|||
Leung
et al. (2001) |
Financial
trading |
|||
Dreger e Schumacher
(2005) |
Economic Indicators /
Germany |
|||
Greer
(2005) |
Interest Rates |
|||
Ramnath
et al. (2005) |
Cash Flow |
|||
Terregrossa
(2005) |
|
|||
Poncela
e Senra (2006) |
Inflation / USA |
|||
Moreno
e López (2007) |
Economic Growth /
Spain |
|||
Hollauer
et al. (2008) |
Brazilian industrial
GDP |
|||
Wang
e Nie (2008) |
Indexes / Shanghai |
|||
Wang
e Nie (2008b) |
Indexes / Shanghai |
|||
Rapach
et al. (2009) |
|
|||
Drechsel e Scheufele
(2011) |
Indicators of industrial production / Germany |
|||
Bjornland
et al. (2012) |
Inflation /
Norway |
|||
Gmez
et al. (2012) |
Food inflation
/ Colombia |
|||
Moreno
e López (2011) |
Macroeconomics
/ Spain |
|||
Industrial demand |
Gardner
(1993) |
Component failure in computer systems |
||
Thomas
(1996) |
Service sector |
|||
Chan
et al. (1999) |
Ordering and stocking of bank printed forms |
|||
Chan
et al. (1999b) |
Inventory management
banking |
|||
Strijbosch
et al. (2000) |
Inventory control procedure for spare parts |
|||
Taylor
et al. (2000) |
Electricity demand |
|||
Cox
e Popken (2002) |
Telecommunications |
|||
Caiado
(2010) |
Water consumption /
Spain |
|||
Lin
et al. (2010) |
Third-generation (3G) telecommunication industry |
|||
Electoral market |
Berg
et al. (2008) |
|
||
Jones
(2008) |
Presidential
elections / USA |
|||
(continuation)
Table 4.
Articles mapped in attribute Application (continuation).
Field |
References |
Context |
||||
Meteorology |
Garand
e Grassotti (1995) |
Rain Rate |
||||
Brown
e Murphy (1996) |
Temperature road
surface |
|||||
Xiong
et al. (2001) |
Precipitation and
runoff |
|||||
Klopper
e Landman (2004) |
South African Weather Service (SAWS) |
|||||
Metzger
et al. (2004) |
El Niño - Southern Oscillation (ENSO) |
|||||
Doblas-Reyes
et al. (2005) |
DEMETER multi-model
ensemble system |
|||||
Lucio
et al. (2007) |
Standardized
Precipitation Index (SPI) |
|||||
Nielsen
et al. (2007) |
Wind energy |
|||||
Bezerra
et al. (2008) |
Climate Forecast |
|||||
Zhang
et al. (2009) |
Hydrological
processes |
|||||
Sumi
et al. (2011) |
Index precipitation |
|||||
He
et al. (2012) |
Hydrological
processes |
|||||
Commercial Sector |
McIntyre
et al. (1993) |
Retailers who plan periodic promotions |
||||
Gong
et al. (2011) |
Foreign trade
forecasting system |
|||||
Mukhopadhyay
et al. (2011) |
Information sharing |
|||||
Zhu
et al. (2011) |
Information sharing |
|||||
Business Sector |
Mahmoud
(1989) |
Managerial issues |
||||
Öller
(1990) |
Business cycle |
|||||
Kamstra
et al. (2001) |
Bond ratings in the
transportation and industrial sectors |
|||||
Jiang
e Yuan (2007) |
Personal credit |
|||||
Yufang
e Minghui (2007) |
Personal credit |
|||||
Rapach
e Strauss (2008) |
Employment growth /
USA |
|||||
Rapach
e Strauss (2012) |
Employment growth /
USA |
|||||
Tourism |
Wong
et al. (2007) |
Tourism demand |
||||
Coshall
(2009) |
Tourism demand |
|||||
Chen
(2011) |
Tourism demand /
Taiwan |
|||||
Other |
Perry
e Euler (1990) |
Situations where time is a scarce resource |
||||
White
et al. (1992) |
Thoroughbred horse
race outcomes |
|||||
Meade
e Islam (1998) |
Technological
forecasting |
|||||
Cho
e Wüthrich (1999) |
Information available on the World Wide Web |
|||||
Host
et al. (2007) |
Laboratory
experiments |
|||||
Stathopoulos
et al. (2008) |
Urban traffic |
|||||
Spann
e Skiera (2009) |
Betting market |
|||||
Zhang
et al. (2010) |
Semiarid mountains |
|||||
Christodoulos
et al. (2011) |
Propagation of a successful innovation |
|||||
Green
e Armstrong (2011) |
Decisions in conflict
situations |
|||||
Most of the articles mapped point at an exploratory
nature. In this class, authors study and analyze various different criteria of
combining forecasts. The 80 selected articles in this attribute were classified
into two Tables, according to the focus of exploration. Table 5 identifies the
studies that explore subjective models, neural networks, Bayesian statistics
and theories for the selection of models. In Table 6 works that focus on the
combined weighting of forecasts, forecast errors and a variety of fields are
listed.
Table 5 presents the scope of subjective methods in
relation to others, despite the bias they may establish (WERNER, 2005). Studies
on the model selection and the use of neurotic networks have also played a major
advance over the last decade. However, the use of Bayesian methods in
combination of predictions is still scarce. Table 6 displays a growing concern
about the weighting of forecasts and the accuracy of the models. Whereas in the
"other" category, still in Table 6, exploration is widespread,
studies address different combining topics, sometimes repeating the context of
antithetical forecasts.
Table 5. Articles mapped in attribute Exploration:
subjective models, neural networks, Bayesian statistics and selection of
models.
Exploration |
||
Subjective Models |
Hogarth (1989) |
Harries et al.
(2004) |
McNees (1992) |
Harvey e Harries (2004) |
|
Sanders e Ritzman (1995) |
Winkler e Clemen (2004) |
|
Maines (1996) |
Richard e Soll (2006) |
|
Vokurka et al. (1996) |
Franses (2011) |
|
Kamstra e Kennedy (1998) |
Franses e Legerstee (2011) |
|
Fischer e Harvey (1999) |
Wang (2011) |
|
Zhou et al. (1999) |
|
|
Neural Networks |
Aussem e Murtagh (1997) |
Szupiluk et al.
(2006) |
Fromm et al. (1998) |
Shi (2009) |
|
Donaldson e Kamstra (1999) |
Aladag et al.
(2010) |
|
Jing-Rong (2000) |
Ranjan e Gneiting (2010) |
|
Magalhaes et al. (2004) |
Wichard (2011) |
|
Bayesian Statistics |
Palm e Zellner (1992) |
Félix e Rodríguez (2008) |
Tibiletti (1994) |
Cai et al.
(2012) |
|
Faria e Souza (1995) |
||
Selection of Models |
Clemen et al. (1995) |
Costantini
e Kunst (2011) |
Zou e Yang (2004) |
Franses (2011b) |
|
Costantini e Pappalardo (2010) |
Table 6. Articles mapped in attribute Exploration:
combined weighting of forecasts, forecast errors and a variety of theme are
listed.
Focus of Exploitation |
||||||||||
Weighting |
Gunter
(1992) |
Theoretical properties |
||||||||
Winkler e Clemen
(1992) |
Instability of
weights |
|||||||||
Chandrasekharan et al. (1994) |
Weights and
covariance matrix |
|||||||||
Mostaghimi
(1996) |
Sensitivity of
weights |
|||||||||
Chan
et al. (2003) |
Variable weights |
|||||||||
Tang
(2003) |
Ideal matrix |
|||||||||
Elliott
e Timmermann (2005) |
Regime change |
|||||||||
Liang
et al. (2006) |
Linear combination |
|||||||||
Fan
e Deng (2007) |
Forecast error to variable weights |
|||||||||
Kim
(2008) |
Generalized
autoregression |
|||||||||
Smith e Wallis (2009) |
Finite-sample error in estimating the combining weights |
|||||||||
Kolassa
(2011) |
Akaike weights |
|||||||||
Errors |
Batchelor
e Dua (1995) |
Expected error
variance |
||||||||
Liu
et al. (1998) |
Distribution of
errors |
|||||||||
Lam
et al. (2001) |
Approaches to minimizing the error |
|||||||||
Wenzel
(2001) |
Alternative measures
to compare |
|||||||||
Tang
(2002) |
Error bounds of optimal combined forecasting (OCF) |
|||||||||
Riedel
(2009) |
Pooling |
|||||||||
Other |
Schmittlein
et al. (1990) |
Winkler method for
combining |
||||||||
Ridley
(1995) |
Global Antithetic
Forecasts |
|||||||||
Gunter
e Aksu (1997) |
Non-negativity Restricted Least Squares – N(E)RLS |
|||||||||
Ridley
(1997) |
Optimal weights for combining antithetic forecasts |
|||||||||
Gardner (1999) |
Rule-based forecasting vs. damped-trend exponential smoothing |
|||||||||
Ridley
(1999) |
Heteroscedastic
antithetic forecasts |
|||||||||
He
e Xu (2005) |
Self-organizing
algorithms |
|||||||||
Preminger
et al. (2007) |
Extended switching
regression model |
|||||||||
Jose
e Winkler (2008) |
Average Trimmed e Winsorized |
|||||||||
Clark
e McCracken (2009) |
Nested models |
|||||||||
Clements
e Harvey (2011) |
Probability forecasts |
|||||||||
Hiernaux
(2011) |
Subspace methods |
|||||||||
Armstrong
(1989) |
Rules to combine |
|||||||||
Ringuest e Tang
(1989) |
Discussion of combinations of Makridakis et al. (1982) |
|
||||||||
Winkler
(1989) |
Forms and rules to combine |
|||||||||
Reeves
e Lawrence (1991) |
Accuracy and direction of change |
|||||||||
Miller
et al. (1992) |
Effects of
nonstationarity |
|||||||||
Holden e Thompson
(1997) |
Combinations, forecast encompassing and efficiency
tests |
|||||||||
Johnston
et al. (1999) |
Series with negligible growth and seasonality |
|||||||||
Terui
e Dijk (2002) |
Nonlinear time series |
|||||||||
Fang
(2003) |
Encompassing tests |
|||||||||
Armstrong
(2007) |
Significance tests |
|||||||||
Wang e Lan (2007) |
Scenario analysis and the technological substitution model |
|
||||||||
Amendola
e Storti (2008) |
Volatility forecasts |
|||||||||
Sancetta
(2009) |
Dependent
heterogeneous data |
|||||||||
Hyndman
et al. (2011) |
Hierarchical time
series |
|||||||||
Theodosiou
(2011) |
STL decomposition |
|||||||||
Besides
the approaches, there is the interest of analyzing the data collected according
to authors. Table 7 lists the number of publications by author, considering all
the authorship of the same work.
According
to Table 7, there are not many papers published by the same author. Of the 321
registered authors, R. L. Winkler leads the ranking with six publications,
followed by R. T. Clemen and J. S. Armstrong, with five publications. As for
the other authors, not included in Table 3, 29 had two publications and 273 had
only one.
Table 7. Number
of publications by author.
Author |
nº |
Author |
nº |
Author |
nº |
Winkler, R. L. |
6 |
Chan, C. K. |
3 |
Ridley, D. |
3 |
Clemen, R. T. |
5 |
Franses, P. H. |
3 |
Shi, Y. |
3 |
Armstrong, J. S. |
5 |
Gunter, S. I. |
3 |
Strauss, J. K. |
3 |
Bunn, D. W. |
4 |
Harvey, N. |
3 |
Taylor, J. W. |
3 |
Kamstra, M. |
4 |
Kingsman, B. G. |
3 |
Wong, H. |
3 |
Jing-Rong, D. |
4 |
Menezes, L. M. |
3 |
Zhou, Z. |
3 |
Rapach, D. E. |
3 |
5. FINAL CONSIDERATIONS
The
use of forecasts has become an important activity in today's market. Regardless
of the mode in which forecasts are obtained, their outcomes affect decision
making. Hence, combining forecasts emerged as a way to gather available
information and increase accuracy of the final forecast. Nevertheless, there
are several methods to combine forecasts. This work presents a literature
review on the approaches of combining forecasts. Aiming to continue the
revision proposed by Clemen (1989), 174 papers published between 1989 and 2012
were collected.
An
overall analysis identified the year 2011 as the peak of the publications of
combining methods, followed by 2007 and 2008 as second in contributions. During
this period Robert L. Winkler stands out with the greatest number of
publications on the subject matter.
Within
the current market framework, combining forecasts is already a widely spread
method in various branches of knowledge.
Economy stands out as the field that invested most in the study of the
subject. In the exploratory approach, however, combining forecast is still a growing
method, as literature proves constant development of new techniques and
comprehensive improvement of previously known models. Among such new
techniques, it is noticeable that, in addition to the artificial neural network
models, different and varied conjectures are proposed for the combination,
along with speculation about the weighting of forecasts and their accuracy
measures. Another highlight is the scarce literature on the use of Bayesian
methods in combining forecasts, indicative of future research.
As
for the comparison scope, many works still confront combining with individual
forecasts, ensuring the accuracy of the models. However, works devoted to
review the matter are still only a few.
The
main contribution of this paper follows the classification and subdivision of
publications in approaches, identifying areas of exploration and the latest
contributions. The analysis of this classification seeks to summarize the
knowledge on the subject so far, and is useful both for researchers in this and
other areas. At last, this work stands as a reference for all those who wish to
combine forecasts.
REFERENCES
Aksu C.; Gunter,
S. I. (1992). An empirical analysis of the accuracy of SA, OLS, ERLS and NRLS
combination forecasts. International
Journal of Forecasting, n. 8, p. 27-43.
ALADAG, C.;
EGRIOGLU, E.; YOLCU, U. (2010). Forecast Combination by Using Artificial Neural
Networks. Neural Processing Letters,
n. 32, p. 269-276.
AMENDOLA, A.;
STORTI G. (2008). A GMM procedure for
combining volatility forecasts. Journal
Computational Statistics & Data Analysis, n. 52, p. 3047-3060.
Armstrong J. S. (1989). Combining forecasts: The end of the beginning
or the beginning of the end? International
Journal of Forecasting, n. 5, p.
585-588.
Armstrong, J. S. (2001). Principles of Forecasting: A Handbook for
Researchers and Practitioners. Kluwer
Academic Publishers.
Armstrong, J. S. (2006). Findings from evidence-based forecasting:
Methods for reducing forecast error. International
Journal of Forecasting, n. 22, p. 583-598.
Armstrong, J. S. (2007). Significance tests harm progress in
forecasting. International Journal of
Forecasting, n. 23, p. 321-327.
AUSSEM, A.;
MURTAGH, F. (1997). Combining Neural Network Forecasts on Wavelet-transformed
Time Series. Connection Science, n. 9,
p. 113-122.
BATCHELOR, R.;
DUA, P. (1995). Forecaster Diversity and the Benefits of Combining Forecasts. Management Science, n. 41, p. 68-75.
Bates, J. M.; Granger, C. W. J. (1969). The
Combining of Forecasts. Operational
Research Quarterly, n. 20, p. 451-468.
BERG, J. E.;
NELSON, F. D.; RIETZ, T. A. (2008). Prediction market accuracy in the long run.
International Journal of Forecasting,
n. 24, p. 285-300.
BEZERRA, A. C.
N.; PEZZI, L. P.; KAYANO, M. T. (2008). Esquema Estatístico de Combinação e Correção de Previsões
Climáticas – ECCOCLIM. Rev.
Bras. Meteorologia, n. 23, p. 347-359.
BJORNLAND, H.
C.; GERDRUP, K.; JORE, A. S.; SMITH, C.; THORSRUD, L. A. (2012). Does Forecast
Combination Improve Norges Bank Inflation Forecast? Oxford Bulletin of Economics and Statistics, n. 74, p. 163-179.
BROWN, B. G.;
MURPHY, A. H. (1996). Improving forecasting performance by combining forecasts:
The example of road-surface temperature forecasts. Meteorological Applications, n. 3, p. 257-265.
CAI, Y.;
STANDER, J.; DAVIES, N. (2012). A new Bayesian approach to quantile
autoregressive time series model estimation and forecasting. Journal of Time Series Analysis, n. 33,
p. 684-698.
CAIADO, J.
(2010) Performance of Combined Double Seasonal Univariate Time Series Models
for Forecasting Water Demand. Journal
of Hydrologic Engineering, n. 15,
p. 215-222.
CASAGRANDE, L. F.; HOSS, O. (2010) Métodos de Forecasting
Conjugado com um Método Qualitativo e um Método com a Média das Previsões
Quantitativas e Qualitativas. CAP –
Accounting and Management, n. 4, p. 94-100.
Chan, C. K.; KINGSMAN, B. G.; WONG, H. (1999). A comparison of unconstrained and constrained
OLS for the combination of demand forecasts: A case study of the ordering and
stocking of bank printed forms. Annals
of Operations Research, n. 87, p. 129-140.
Chan, C. K.; Kingsman, B.
G.; Wong, H. (1999b). The
value of combining forecasts in inventory management - a case study in banking.
European Journal of Operational Research,
n. 117, p. 199-210.
Chan, C. K.; Kingsman, B.
G.; Wong, H. (2003).
Determining when to update the weights in combined forecasts for product
demand--an application of the CUSUM technique. European Journal of Operational Research, n. 153, p. 757-768.
CHANDRASEKHARAN,
R.; MORIARTY, M. M.; WRIGHT, G. P. (1994). Testing for unreliable estimators
and insignificant forecasts in combined forecasts. Journal of Forecasting, n. 13, p. 611-624.
CHEN, K. Y.
(2011). Combining linear and nonlinear model in forecasting tourism demand. Expert Systems with Applications: An
International Journal, n. 38, p. 10368-10376.
CHO, V.;
WÜTHRICH, B. (1999). Combining Forecasts from Multiple Textual Data Sources. Methodologies for Knowledge Discovery and
Data Mining, n. 1574, p. 174-179.
CHRISTODOULOS,
C.; MICHALAKELIS, C.; VAROUTAS, D. (2011). On the combination of exponential
smoothing and diffusion forecasts: An application to broadband diffusion in the
OECD area. Technological Forecasting and
Social Change, n. 78, p. 163-170.
CLARK, T. E.;
MCCRACKEN, M. W. (2009). Combining Forecasts from Nested Models†. Oxford Bulletin of Economics and Statistics,
n. 71, p. 303-329.
Clemen, R. T. (1989). Combining forecasts: A review and annotated
bibliography. International Journal of
Forecasting, n. 5, p. 559-583.
Clemen, R. T.; MURPHY, A. H.; WINKLER, R. L. (1995). Screening
probability forecasts: contrasts between choosing and combining. International Journal of Forecasting,
n. 11, p. 133-145.
CLEMENTS, M. P.;
HARVEY, D. I. (2011). Combining probability forecasts. International Journal of Forecasting, n. 27, p. 208-223.
COLLOPY, F.;
ARMSTRONG, J. S. (1992).
Rule-based forecasting: development and validation of an expert systems
approach to combining time series extrapolations. Journal Management Science, n. 38, p. 1394-1414.
COSHALL, J. T. (2009).
Combining volatility and smoothing forecasts of UK demand for international
tourism. Tourism Management, n. 30,
p. 495-511.
COSTANTINI, M.;
PAPPALARDO, C. (2010) A hierarchical procedure for the combination of
forecasts. International Journal of Forecasting,
n. 26, p. 725-743.
COSTANTINI, M.;
KUNST, R. M. (2011). Combining forecasts based on multiple encompassing tests
in a macroeconomic core system. Journal
of Forecasting, n. 30, p. 579-596.
COX, L. A. Jr.;
POPKEN, D. A. (2002). A hybrid system-identification method for forecasting
telecommunications product demands. International
Journal of Forecasting, n. 18, p. 647-671.
DOBLAS-REYES, F.
J.; HAGEDORN, R.; PALMER, T. N. (2005). The rationale behind the success of
multi-model ensembles in seasonal forecasting – II: Calibration and
combination. Tellus A, n. 57, p.
234-252.
Donaldson, R. G.; Kamstra, M. (1996). Forecast combining with
neural networks. Journal of Forecasting,
n. 15, p. 49-61.
Donaldson, R. G.; Kamstra, M. (1999). Neural network forecast
combining with interaction effects. Journal
of the Franklin Institute, n. 336, p. 227-236.
Drechsel, K.; Scheufele, R. (2012). The performance of
short-term forecasts of the German economy before and during the 2008/2009
recession. International Journal of
Forecasting, n. 28, p. 428–445.
DREGER, C.;
SCHUMACHER, C. (2005). Out-of-sample Performance of Leading Indicators for the
German Business Cycle: Single vs. Combined Forecasts. Journal of Business Cycle Measurement and Analysis 2005: p. 71-87.
ELLIOTT, G.;
TIMMERMANN, A. (2005). Optimal Forecast Combination Under Regime Switching. International Economic Review, n. 46,
p. 1081-1102.
FAN, W. G.;
DENG, F. (2007). Variable Weight Combining Forecasts Based on Forecasting
Error. International Conference on Control
and Automation, n. 5, p. 1610-1613.
FANG ,Y. (2003).
Forecasting combination and encompassing tests. International Journal of Forecasting, n. 19, p. 87-94.
FARIA, A. E.;
SOUZA, R. C. (1995). A re-evaluation of the quasi-bayes approach to the linear
combination of forecasts. Journal of
Forecasting, n. 14, p. 533-542.
FÉLIX, J. A.;
RODRÍGUEZ, F. F. (2008). Improving moving average trading rules with boosting
and statistical learning methods. Journal
of Forecasting, n. 27, p. 433-449.
FISCHER, I.;
HARVEY, N. (1999). Combining forecasts: What information do judges need to
outperform the simple average? International
Journal of Forecasting, n. 15, p. 227-246.
FRANSES, P. H.
(2011). Averaging Model Forecasts and Expert Forecasts: Why Does It Work? Journal Interfaces, n. 41, p. 177-181.
FRANSES, P. H.
(2011b). Model selection for forecast combination. Applied Economics, n. 43, p. 1721-1727.
FRANSES, P. H.;
LEGERSTEE, R. (2011). Combining SKU - level sales forecasts from models and
experts. Expert Systems with Applications,
n. 38, p. 2365-2370.
FROMM, J.; FEI,
H.; FIORDALISO, A. (1998). A nonlinear forecasts combination method based on
Takagi-Sugeno fuzzy systems. International
Journal of Forecasting, n. 14, p. 367-379.
GARAND, L.;
GRASSOTTI, C. (1995). Toward an objective analysis of rainfall rate combining
observations and short-term forecast model estimates. Journal of Applied Meteorology, n. 34, p. 1962-77.
GARDNER, E. S.
Jr. (1993). Forecasting the failure of component parts in computer systems: A
case study. International Journal of
Forecasting, n. 9, p. 245-253.
GARDNER, E. S.
Jr. (1999). Note: Rule-Based Forecasting Vs. Damped-Trend Exponential
Smoothing. Journal Management Science,
n. 45, p. 1169-1176.
GMEZ, M. I.; GONZLEZ, E. R.; MELO, L. F. (2012). Forecasting
Food Inflation in Developing Countries with Inflation Targeting Regimes. American Journal of Agricultural Economics,
n. 94, p. 153-173.
GONG, K.; LIU,
M.; FANG, Y.; ZHANG, X. (2011). A Hybrid Model of Rough Sets and Shannon
Entropy for Building a Foreign Trade Forecasting System. Computational Sciences and
Optimization: p. 7-11.
GOODWIN, P.
(2000). Correct or combine? Mechanically integrating judgmental forecasts with
statistical methods. International
Journal of Forecasting, n. 16, p. 261-275.
GREEN, K. C.; Armstrong, J. S. (2011). Role thinking:
Standing in other people’s shoes to forecast decisions in conflicts. International Journal of Forecasting,
n. 27, p. 69-80.
GREER, M.
(2005). Combination forecasting for directional accuracy: An application to
survey interest rate forecasts. Journal
of Applied Statistics, n. 32, p. 607-615.
Gunter, S. I. (1992). Nonnegativity restricted least squares
combinations. International Journal of
Forecasting, n. 8, p. 45-59.
Gunter, S. I.; Aksu
C. (1997). The usefulness of heuristic N(E)RLS algorithms for combining
forecasts. Journal of Forecasting,
n. 16, p. 439-462.
HARRIES, C.;
YANIV, I.; HARVEY, N. (2004). Combining advice: the weight of a dissenting
opinion in the consensus. Journal of
Behavioral Decision Making, n. 17, p. 333-348.
HARVEY, N.;
HARRIES, C. (2004). Effects of judges' forecasting on their later combination
of forecasts for the same outcomes. International
Journal of Forecasting, n. 20, p. 391-409.
HE, C.; XU, X.
(2005). Combination of forecasts using self-organizing algorithms. Journal of Forecasting, n. 24, p.
269-278.
HE, Z.; ZHAO,
W.; LIU, H. (2012). Comparing the performance of
empirical black-box models for river flow forecasting in the Heihe River Basin,
Northwestern China. Hydrological
Processes (in press).
HIERNAUX, A. G.
(2011). Forecasting linear dynamical systems using subspace methods. Journal of Time Series Analysis, n. 32,
p. 462-468.
HOGARTH, R. M.
(1989). On combining diagnostic
‘forecasts’: Thoughts and some evidence. International
Journal of Forecasting, n. 5, p. 593-597.
HOLDEN, K.;
THOMPSON, J. (1997). Combining forecasts, encompassing and the properties of UK
macroeconomic forecasts. Applied
Economics, n. 29, p. 1447-1458.
HOLLAUER,
G.; ISSLER, J. V.; NOTINI, H. H. (2008). Prevendo o crescimento da produção industrial
usando um número limitado de combinações de previsões. Economia Aplicada, n. 12, p. 177-198.
HOST, O.; LAHAV, O.; ABDALLA, F. B.; EITEL, K. (2007).
Forecasting neutrino masses from combining KATRIN and the CMB observations:
Frequentist and Bayesian analyses. Physical
Review D76: 113005.
HSIAO, C.; WAN,
S. K. (2011). Comparison of forecasting methods with an application to
predicting excess equity premium. Mathematics
and Computers in Simulation, n. 81, p. 1235-1246.
HU, M. Y.;
TSOUKALAS, C. (1999). Combining conditional volatility forecasts using neural
networks: an application to the EMS exchange rates. Journal
of International Financial Markets, n. 9, p. 407-422.
HYNDMAN, R. J.;
AHMED, R. A.; ATHANASOPOULOS G.; SHANG HL. (2011). Optimal combination
forecasts for hierarchical time series. Journal
Computational Statistics & Data Analysis, n. 55, p. 2579-2589.
JEONG, D.; KIM,
Y. (2009). Combining single-value stream flow forecasts – A review and
guidelines for selecting techniques. Journal
of Hydrology, n. 377, p. 284-299.
JIANG, M.; YUAN,
X. (2007). Personal Credit Scoring Model of Non-linear Combining Forecast Based
on GP. Third International Conference on
Natural Computation, n. 4, p. 408-414.
JING-RONG, D.
(2000). Research on the method of nonlinear combining forecasts based on
fuzzy-neural systems. Third World
Congress on Intelligent Control and Automation, n. 2, p. 899-903.
JING-RONG, D.
(2002). A nonlinear combining forecast method based on fuzzy neural network. International Conference on Machine
Learning and Cybernetics, n. 4,
p. 2160-2164.
JING-RONG, D.
(2007). Combining Stock Market Volatility Forecasts Using an EWMA Technique. International Conference on Wireless
Communications, Networking and Mobile Computing (WiCom): p. 5277-5280.
JING-RONG, D.
(2008). Combining Stock Market Volatility Forecasts Using a Bayesian Technique.
International Conference on Wireless
Communications, Networking and Mobile Computing (WiCom): p. 1-5.
JOHNSTON, F. R.;
BOYLAN, J. E.; SHALE, E.; MEADOWS, M. (1999). A robust forecasting system,
based on the combination of two simple moving averages. Journal of the Operational Research Society, n. 50, p. 1199-1204.
JONES, R. J. Jr.
(2008). The state of presidential election forecasting: The 2004 experience. International Journal of Forecasting,
n. 24, p. 310-321.
JORGENSEN, M.
(2007). Forecasting of software development work effort: Evidence on expert
judgement and formal models. International
Journal of Forecasting, n. 23, p. 449-462.
JOSE, V. R. R.;
WINKLER, R. L. (2008). Simple robust averages of forecasts: Some empirical
results. International Journal of
Forecasting, n. 24, p. 163-169.
KAMSTRA, M.;
KENNEDY, P. (1998). Combining qualitative forecasts using logit. International Journal of Forecasting,
n. 14, p. 83-93.
KAMSTRA, M.;
KENNEDY, P.; SUAN, T. K. (2001). Combining Bond Rating Forecasts Using Logit. Financial Review, n. 36, p. 75-96.
KIM, J. R. K.
(2008). Combining forecasts using optimal combination weight and generalized
autoregression†. Journal of Forecasting,
n. 27, p. 419-432.
KLEIN, L. R.;
PARK, J. Y. (1993). Economic forecasting at high-frequency intervals. Journal of Forecasting, n. 12, p.
301-319.
KLOPPER, E.;
LANDMAN, W. (2004). A simple approach for combining seasonal forecasts for
southern Africa. Meteorological
Applications, n. 10, p. 319-327.
KOLASSA, S.
(2011). Combining exponential smoothing forecasts using Akaike weights. International Journal of Forecasting,
n. 27, p. 238-251.
LAM, K. F.; MUI,
H. W.; YUEN, H. K. (2001). A note on minimizing absolute percentage error in
combined forecasts. Computers and
Operations Research, n. 28, p. 1141-1147.
LARRICK, R. P.;
SOLL, J. B. (2006). Intuitions About Combining Opinions: Misappreciation of the
Averaging Principle. Journal Management
Science, n. 52, p. 111-127.
LEUNG, M. T.;
DAOUK, H.; CHEN, A. (2001). Using investment portfolio return to combine
forecasts: A multi objective approach. European
Journal of Operational Research, n. 134, p. 84-102.
LIANG, K. Y.;
LEE, J. C.; SHAO, K. S. H. (2006). On the Distribution of the Inverted Linear
Compound of Dependent F-Variates and its Application to the Combination of
Forecasts. Journal of Applied Statistics,
n. 33, p. 961-973.
LIN, C. C.;
TANG, Y. H.; SHYU, J. Z.; LI, Y. M. (2010) Combining forecasts for technology
forecasting and decision making. Journal
of Technology Management in China, n. 5, p. 69-83.
LIU, J.; KRENZ,
D. C.; GALVEZ, A. F.; LUMEN, B. O.; MENEZES, L. M.; BUNN, D. W. (1998). The
persistence of specification problems in the distribution of combined forecast
errors. International Journal of
Forecasting, n. 14, p. 415-426.
LOBO, G. J.;
NAIR, R. D. (1990). Combining Judgmental and Statistical Forecasts: An
Application to Earnings Forecasts. Decision
Sciences, n. 21, p. 446-460.
LOBO, G. J.
1991. Alternative methods of combining security analysts' and statistical
forecasts of annual corporate earnings. International
Journal of Forecasting, n. 7, p. 57-63.
LUBECKE, T. H.;
NAM, K. D.; MARKLAND, R. E.; KWOK, C. C. Y. (1998). Combining foreign exchange
rate forecasts using neural networks. Global
Finance Journal, n. 9, p. 5-27.
LUCIO, P. S.; SANTOS, L. A.; SILVA, F. D.; BALBINO, H. T.;
FERREIRA, D. B.; SALVADOR, M. A. (2007). Combining
stochastic forecasts of attributes based on the Standardised Precipitation
Index transformation design. Geophysical
Research Abstracts, n. 9, p. 10266.
MACDONALD, R.;
MARSH, I. W. (1994). Combining exchange rate forecasts: What is the optimal
consensus measure? Journal of
Forecasting, n. 13, p. 313-332.
MAGALHAES, M.
H.; BALLINI, R.; MOLCK, P.; GOMIDE, F. (2004). Combining forecasts for natural
stream flow prediction. Annual Meeting
of the Fuzzy Information, n. 1, p. 390-394.
MAHMOUD, E.
(1989). Combining forecasts: Some managerial issues. International Journal of Forecasting, n. 5, p. 599-600.
MAINES, L. A.
(1996). An experimental examination of
subjective forecast combination. International
Journal of Forecasting, n. 12, p. 1178-1195.
MAKRIDAKIS, S.;
ANDERSON, A.; CARBONE, R.; FILDES, R.; HIBON, M.; LEWANDOSKI, R.; NEWTON, J.;
PARZEN, E.; WINKLER, R. (1982). The accuracy of extrapolation (time series)
methods: Results of a forecasting competition. Journal of Forecasting, n. 1, p. 111-153.
MAKRIDAKIS, S.;
CHATFIELD, C.; HIBON, M.; LAWRENCE, M.; MILLS, T.; ORD, K.; SIMMONS, L. F.
(1993). The M2-Competition: a real-time judgmentally based forecasting study. International Journal of Forecasting,
n. 9, p. 5-22.
MAKRIDAKIS, S.;
HIBON, M. (2000). The M3-Competition: results, conclusions and implications. International Journal of Forecasting, n. 16, p. 451-476.
MARQUES, E. B. (2005). Combinação
de Previsões de Índices de Preços. Dissertation (Master in Economy). Rio de Janeiro: Fundação Getúlio
Vargas.
MARTINS, V. L. M. (2011). Comparação de combinação de previsões correlacionadas e não
correlacionadas com as suas previsões individuais: um estudo com séries
industriais. Dissertation (Master in Production Engineering). Porto
Alegre: PPGEP/UFRGS.
MARTINS, V. L.
M.; WERNER, L. (2012). Forecast combination in industrial series: A comparison
between individual forecasts and its combinations with and without correlated
errors. Expert Systems with Applications,
n. 39, p. 11479-11486.
MCINTYRE, S. H.;
ACHABAL, D. D.; MILLER, C. M. (1993). Applying case-based reasoning to forecasting
retail sales. Journal of Retailing, n.
69, p. 372-398.
MCNEES, S. K.
(1992). The uses and abuses of ‘consensus’ forecasts. Journal of Forecasting, n. 11, p. 703-710.
MEADE,
N.; ISLAM, T. (1998). Technological Forecasting - Model Selection, Model Stability,
and Combining Models. Management Science, n. 44, p. 1115-1130.
MENEZES, L. M.;
BUNN, D. W. (1993). Diagnostic tracking and model specification in combined
forecasts of U.K. inflation. Journal of
Forecasting, n. 12, p. 559-572.
MENEZES, L. M.;
BUNN, D. W.; TAYLOR, J. W. (2000). Review of guidelines for the use of combined
forecasts. European Journal of
Operational Research, n. 120, p. 190-204.
METZGER, S.;
LATIF, M.; FRAEDRICH, K. (2004). Combining ENSO Forecasts: A Feasibility
Study. Monthly Weather Review, n. 132, p. 456-72.
MILLER, C. M.;
CLEMEN, R. T.; WINKLER, R. L. (1992). The effect of nonstationarity on combined
forecasts. International Journal of
Forecasting, n. 7, p. 515-529.
SHI, S. M.; XU,
L. D.; LIU, B. (1999). Improving the accuracy of nonlinear combined forecasting
using neural networks. Expert Systems
with Applications, n. 16, p. 49-54.
MORENO, B.;
LÓPEZ, A. J. (2007). Combining economic forecasts through information measures.
Applied Economics Letters, n. 14, p.
899-903.
MORENO, B.; LÓPEZ,
A. J. (2011). Combining Economic Forecasts by Using a Maximum Entropy
Econometric Approach. Journal of
Forecasting (in press).
MOSTAGHIMI, M.
(1996). Combining ranked mean value forecasts. European Journal of Operational Research, n. 94, p. 505-516.
MUKHOPADHYAY, S.
K.; YUE, X.; ZHU, X. (2011). A Stackelberg model of pricing of complementary
goods under information asymmetry. International
Journal of Production Economics, n. 134, p. 424-433.
NEWBOLD, P.;
GRANGER, C. W. J. (1974).
Experience with Forecasting Univariate Time Series and the Combination of
Forecasts. Journal of Royal Statistical
Society (A), n. 137, p. 131-165.
NIELSEN, H. A.;
NIELSEN, T. S.; MADSEN, H.; PINDADO, M. J. S. I.; MARTI, I. (2007). Optimal
combination of wind power forecasts. Wind
Energy, n. 10, p. 471-482.
ÖLLER,
L. E. (1990). Forecasting the business cycle using survey data. International Journal of Forecasting, n. 6,
p. 453-461.
PALM, F. C.;
ZELLNER, A. (1992). To combine or not to combine? Issues of combining
forecasts. Journal of Forecasting,
n. 11, p. 687-701.
PERRY, C.;
EULER, T. (1990). Cost-effective forecasting: Lessons my computer programs
never taught me. Omega, n. 18, p.
241-246.
PALIWAL, M.;
KUMAR, U. (2009). A Neural networks and statistical techniques: A review of
applications. Expert Systems with
Applications, n. 36, p. 2-17.
PONCELA, P.;
SENRA, E. (2006). A two factor model to combine US inflation forecasts. Applied Economics, n. 38, p. 2191-2197.
PONCELA, P.; RODRÍGUEZ, J.; MANGAS, R. S.; SENRA, E.
(2011). Forecast combination through dimension reduction
techniques. International Journal of
Forecasting, n. 27, p. 224-237.
PREMINGER, A.;
ZION, U. B.; WETTSTEIN, D. (2007). The extended switching regression model:
allowing for multiple latent state variables. Journal of Forecasting, n. 26, p. 457-473.
RAMNATH, S.;
ROCK, S.; SHANE, P. (2005). Value Line and I/B/E/S earnings forecasts .International Journal of Forecasting,
n. 21, p. 185-198.
RANJAN, R.;
GNEITING, T. (2010) Combining probability forecasts. Journal of the Royal Statistical Society, n. 72, p. 71-91.
RAPACH, D. E.;
STRAUSS, J. K. (2008). Forecasting US employment growth using forecast
combining methods. Journal of
Forecasting, n. 27, p. 75-93.
RAPACH, D. E.;
STRAUSS, J. K.; ZHOU, G. (2009). Out-of-Sample Equity Premium Prediction:
Combination Forecasts and Links to the Real Economy. Review of Financial Studies, n. 23, p. 821-862(42).
RAPACH, D. E.;
STRAUSS, J. K. (2012). Forecasting US state-level employment growth: An
amalgamation approach. International Journal
of Forecasting, n. 28, p. 315-327.
REEVES, G. R.;
LAWRENCE, K. D. (1991). Combining forecasts given different types of
objectives. European Journal of
Operational Research, n. 51, p. 65-72.
RIDLEY, D.
(1995). Combining Global Antithetic Forecasts. International Transactions in Operational Research, n. 2, p.
387-398.
RIDLEY, D.
(1997). Optimal weights for combining antithetic forecasts. Computers and Industrial Engineering,
n. 32, p. 371-381.
RIDLEY, D.
(1999). Combining heteroscedastic antithetic forecasts. Computers and Industrial Engineering, n. 36, p. 227-230.
RIEDEL, S.
(2009). Pooling for Combination of Multilevel Forecasts. Transactions on Knowledge and Data Engineering, n. 21, p. 1753-66.
RINGUEST, J. L.;
TANG, K. (1989). An empirical comparison of five procedures for combining (or
selecting) forecasts. Socio-Economic
Planning Sciences, n. 23, p. 217-225.
SANCETTA, A.
(2009). Recursive Forecast Combination for Dependent Heterogeneous Data. Econometric Theory, n. 26, p. 598-631.
SANDERS, N. R.;
RITZMAN, L. P. (1995). Bringing judgment into combination forecasts. Journal of Operations Management, n. 13,
p. 311-321.
SCHMITTLEIN, D.
C.; KIM, J.; MORRISON, D. G. (1990). Combining forecasts: operational
adjustments to theoretically optimal rules. Journal Management Science, n. 36, p. 1044-1056.
SHEN, C. H.
(1996). Forecasting macroeconomic variables using data of different
periodicities. International Journal of
Forecasting, n. 12, p. 269-282.
SHEN, S.; LI,
G.; SONG, H. (2008). An Assessment of Combining Tourism Demand Forecasts over
Different Time Horizons. Journal of
Travel Research, n. 47, p. 197-207.
SHI, S. Y.
(2009). Study on the Technique and Error Analysis of Nonlinear Combining
Forecasts Based Fuzzy System. International
Conference on Electronic Computer Technology: p. 385-387.
SILVA, E. L.;
MENEZES, E. M. (2001). Metodologia da Pesquisa e
Elaboração de Dissertação. Laboratório
de Ensino à Distância da UFSC / PPGEP: 3ª Ed.
SMITH, S. K.
(1997). Further thoughts on simplicity and complexity in population projection
models. International Journal of
Forecasting, n. 13, p. 557-565.
SMITH, J.;
WALLIS, K. F. (2009). A Simple Explanation of the Forecast Combination Puzzle. Oxford Bulletin of Economics and Statistics,
n. 71, p. 331-355.
SPANN, M.;
SKIERA, B. (2009). Sports forecasting: a comparison of the forecast accuracy of
prediction markets, betting odds and tipsters. Journal of Forecasting, n. 28, p. 55-72.
SPIRO, P. S.
(1989). Improving a group forecast by removing the conservative bias in its
components. International Journal of
Forecasting, n. 5, p. 127-131.
STATHOPOULOS,
A.; DIMITRIOU, L.; TSEKERIS, T. (2008). Fuzzy Modeling Approach for Combined
Forecasting of Urban Traffic Flow. Computer-Aided Civil and Infrastructure
Engineering, n. 23, p. 521-535.
STRIJBOSCH, L.
W. G.; HEUTS, R. M. J.; VAN DER SCHOOT, E. H. M. (2000). A combined forecast -
inventory control procedure for spare parts. Journal of the Operational Research Society, n. 51, p. 1184-1192.
SUMI, S. M.;
ZAMAN, F.; HIROSE, H. (2011). A novel hybrid forecast model with weighted
forecast combination with application to daily rainfall forecast of Fukuoka
city. ACIIDS’11 Proceedings of the Third
international conference on Intelligent information and database systems: p.
262-271.
SZUPILUK, R.;
WOJEWNIK, P.; ZABKOWSKI, T. (2006). Combining forecasts with blind signal
separation methods in electric load prediction framework. 24th IASTED international conference on Artificial intelligence and
applications table of contents: p. 439-444.
TANG, X.; ZHOU, Z.; SHI, Y. (2002). The
error bounds of combined forecasting. Mathematical
and Computer Modeling, n. 36, p. 997-1005.
TANG, X.; ZHOU,
Z.; SHI, Y. (2003). The variable weighted functions of combined forecasting. Computers and Mathematics with Applications,
n. 45, p. 723-730.
TAYLOR, J. W.;
BUNN, D. W. (1999). Investigating Improvements in the Accuracy of Prediction
Intervals for Combinations of Forecasts: A Simulation Study. International Journal of Forecasting,
n. 15, p. 325-339.
TAYLOR, J. W.;
MAJITHIA, S.; JAMES, W. (2000). Using combined forecasts with changing weights
for electricity demand profiling. Journal
of the Operational Research Society, n. 51, p. 72-82.
TERREGROSSA, S.
J. (2005). On the efficacy of constraints on the linear combination forecast
model. Applied Economics Letters, n.
12, p. 19-28.
TERUI, N.; DIJK,
H. K. (2002). Combined forecasts from linear and nonlinear time series models. International Journal of Forecasting, n. 18, p. 421-438.
THEODOSIOU, M.
(2011). Forecasting monthly and quarterly time series using STL decomposition. International Journal of Forecasting,
n. 27, p. 1178-1195.
THOMAS, R. J.
(1996). Estimating demand for services: issues in combining sales forecasts. Journal of Retailing and Consumer Services,
n. 3, p. 241-250.
TIBILETTI, L.
(1994). A non-linear combination of experts' forecasts: A Bayesian approach. Journal of Forecasting, n. 13, p.
21-27.
TSANGARI, H.
(2007). An Alternative Methodology for Combining Different Forecasting Models. Journal of Applied Statistics, n. 34,
p. 403-421(19).
VOLKOV, V. Y.;
GLADKOV, Y. U. M. (1995). Reconfigurable combined forecasts in a non-stationary
inflationary environment. Journal of
Forecasting, n. 14, p. 395-403.
VOKURKA, R. J.;
FLORES, B. E.; PEARCE, S. L. (1996). Automatic feature identification and
graphical support in rule-based forecasting: a comparison. International Journal of Forecasting, n. 12, p. 495-512.
WALLIS, K.
(2011). Combining forecasts - forty years later. Applied Financial Economics, n. 21, p. 33-41.
WALZ, D. B.; WALZ, D. P. (1989). Combining
Forecasts: Multiple Regression versus a Bayesian Approach†. Decision Sciences, n. 20, p. 77-89.
WANG, M. Y.;
LAN, W. T. (2007). Combined forecast process: Combining scenario analysis with
the technological substitution model. Technological
Forecasting and Social Change, n.
74, p. 357-378.
WANG, W.; NIE,
S. (2008). The Performance Evaluation and Choice of Combining Forecast Method. Second International Symposium on
Intelligent Information Technology Application, n. 1, p. 838-842.
WANG, W.; NIE,
S. (2008b). The Performance of Several Combining Forecasts for Stock Index. International Seminar on Future Information
Technology and Management Engineering: p. 450-455.
WANG, G.;
KULKARNI, S. R.; POOR, H. V.; OSHERSON, D. N. (2011). Aggregating Large Sets of
Probabilistic Forecasts by Weighted Coherent Adjustment. Journal Decision Analysis, n. 8, p. 128-144.
WEBBY, R.;
O’CONNOR, M. (1996). Judgment and Statistical Time Series Forecasting: a Review
of the Literature. International Journal
of Forecasting, n. 12, p. 91-118.
WENZEL, T.
(2001). Hits-and-misses for the evaluation and combination of forecasts. Journal of Applied Statistics, n. 28, p.
759-773.
WERNER, L.
(2005). Um modelo composto para realizar
previsão de demanda através da integração da combinação de previsões e do
ajuste baseado na opinião. Thesi (PhD in Production Engineering). Porto Alegre:
UFRGS.
WERNER, L.;
RIBEIRO, J. L. D. (2006). Modelo
Composto para prever demanda através da integração de previsões. Produção (São Paulo), n. 16, p. 493-509.
WHITE, E. M.; DATTERO, R.; FLORES, B. (1992). Combining
vector forecasts to predict thoroughbred horse race outcomes. International Journal of Forecasting,
n. 8, p. 595-611.
WICHARD, J. D.
(2011). Forecasting the NN5 time series with hybrid models. International Journal of Forecasting,
n. 27, p. 700-707.
WINKLER, R. L.
(1989). Combining forecasts: A philosophical basis and some current issues. International Journal of Forecasting,
n. 5, p. 605-609.
WINKLER, R. L.; Clemen,
R. T. (1992). Sensitivity of weights in combining forecasts. Operations Research, n. 40, p. 609-614.
WINKLER, R. L.;
CLEMEN, R. T. (2004). Multiple Experts vs. Multiple Methods: Combining
Correlation Assessments. Journal
Decision Analysis, n. 1, p. 167-176.
WONG, K. K. F.;
SONG, H.; WITT, S. F.; WU, D. C. (2007). Tourism forecasting: To combine or not
to combine? Tourism Management, n. 28,
p. 1068-1078.
XIONG, L.;
SHAMSELDIN, A. Y.; O’CONNOR, K. M. (2001). A non-linear combination of the
forecasts of rainfall-runoff models by the first-order Takagi-Sugeno fuzzy
system. Journal of Hydrology, n. 25,
p. 196-217.
YUFANG, C.;
MINGHUI, J. (2007). Combining Forecasts of Personal Credit Scoring Based on RBF
Neural Network. Control Conference
(Chinese): p. 250-252.
ZHANG, L.; ZHAO,
W.; HE, Z.; LIU, H. (2009). Application of the Takagi - Sugeno fuzzy system for
combination forecasting of river flow in semiarid mountain regions. Hydrological Processes, n. 23, p.
1430-1436.
ZHANG, X.; LEI,
Y.; CAO, Q. V. (2010). Compatibility of Stand Basal Area Predictions Based on
Forecast Combination. Forest Science, n. 56, p. 552-557.
ZHOU, Z.; SHI,
Y.; HAO, X. (1999). An MC2 Linear Programming Approach to Combined Forecasting.
Mathematical and Computer Modeling,
n. 29, p. 97-103.
ZHU, X.; MUKHOPADHYAY,
S. K.; YUE, X. (2011). Role of forecast effort on supply chain profitability
under various information sharing scenarios. International Journal of Production Economics, n. 129, p. 284-291.
ZOU, H.; YANG,
Y. (2004). Combining time series models for forecasting. Journal of Forecasting, n. 20,
p. 69-84.