Rough Set Theory

3
Rough Set Theory – Fundamental Concepts,

Principals, Data Extraction, and Applications
Silvia Rissino1 and Germano Lambert-Torres2
1Federal University of Rondonia,
2ItajubaFederal University
Brazil
1. Introduction
Rough Set Theory, proposed in 1982 by Zdzislaw Pawlak, is in a state of constant
development. Its methodology is concerned with the classification and analysis of imprecise,
uncertain or incomplete information and knowledge, and of is considered one of the first
non-statistical approaches in data analysis (Pawlak, 1982).
The fundamental concept behind Rough Set Theory is the approximation of lower and
upper spaces of a set, the approximation of spaces being the formal classification of
knowledge regarding the interest domain.
The subset generated by lower approximations is characterized by objects that will
definitely form part of an interest subset, whereas the upper approximation is characterized
by objects that will possibly form part of an interest subset. Every subset defined through
upper and lower approximation is known as Rough Set.
Over the years Rough Set Theory has become a valuable tool in the resolution of various
problems, such as: representation of uncertain or imprecise knowledge; knowledge analysis;
Open Access Database www.intechweb.org
evaluation of quality and availability of information with respect to consistency and

presence a not of date patterns; identification and evaluation of date dependency; reasoning
based an uncertain and reduct of information data.
The extent of rough set applications used today is much wider than in the past, principally
in the areas of medicine, analysis of database attributes and process control. The subject of
this chapter is to present the Rough Set Theory, important concepts, and Rough Set Theory
used with tools for data mining, special applications in analysis of data in dengue diagnosis.
•
The chapter is divided into the four following topics:
•
Fundamental concepts
•
Rough set with tools for data mining
•
Applications of rough set theory;
Case – Rough set with tools in dengue diagnosis.
2. Fundamental concepts
Rough Sets Theory has been under continuous development for over years, and a growing
number of researchers have became its interested in methodology. It is a formal theory
derived from fundamental research on logical properties of information systems. From the
Source: Data Mining and Knowledge Discovery in Real Life Applications, Book edited by: Julio Ponce and Adem Karahoca,
ISBN 978-3-902613-53-0, pp. 438, February 2009, I-Tech, Vienna, Austria
www.intechopen.com
36 Data Mining and Knowledge Discovery in Real Life Applications
outset, rough set theory has been a methodology of database mining or knowledge
discovery in relational databases. This section presents the concepts of Rough Set Theory;
which coincide partly with the concepts of other theories that treat uncertain and vagueness
information. Among the existent, most traditional approaches for the modeling and
treatment of uncertainties, they are the Theories of the Uncertainty of Dempster-Shafer and
Fuzzy Set (Pawlak et al., 1995). The main concepts related to Rough Set Theory are
presented as the following:
2.1 Set
A set of objects that possesses similar characteristics it is a fundamental part of mathematics.
All the mathematical objects, such as relations, functions and numbers can be considered as
a set. However, the concept of the classical set within mathematics is contradictory; since a
set is considered to be "grouping" without all elements are absent and is know as an empty
set (Stoll, 1979). The various components of a set are known as elements, and relationship
between an element and a set is called of a pertinence relation. Cardinality is the way of
measuring the number of elements of a set. Examples of specific sets that treat vague and
imprecise date are described below:
a. Fuzzy Set
Proposed by mathematician Loft Zadeh in the second half of the sixties, it has as its objective
the treatment of the mathematical concept of vague and approximate, for subsequent
programming and storage on computers.
In order for Zadeh to obtain the mathematical formalism for fuzzy set, it was necessary to
use the classic set theory, where any set can be characterized by a function. In the case of the
fuzzy set, the characteristic function can be generalized so that the values are designated as
elements of the Universe Set U belong to the interval of real numbers [0,1].
The characteristic Function Fuzzy is µA: U å [0,1], where the values indicate the degree of
pertinence of the elements of set U in relation to the set A, which indicated as it is possible
for an element of x of U to belong to A, this function is known as Function of Pertinence and
the set A is the Fuzzy Set (Zadeh, 1965).
b. Rough Set
An approach first forwarded by mathematician Zdzislaw Pawlak at the beginning of the
eighties; it is used as a mathematical tool to treat the vague and the imprecise. Rough Set
Theory is similar to Fuzzy Set Theory, however the uncertain and imprecision in this
approach is expressed by a boundary region of a set, and not by a partial membership as in
Fuzzy Set Theory. Rough Set concept can be defined quite generally by means of interior
and closure topological operations know approximations (Pawlak, 1982).
Observation:
It is interesting to compare definitions of classical sets, fuzzy sets and rough sets. Classical
set is a primitive notion and is defined intuitively or axiomatically. Fuzzy set is defined by
employing the fuzzy membership function, which involves advanced mathematical
structures, numbers and functions. Rough set is defined by topological operations called
approximations, thus this definition also requires advanced mathematical concepts.
2.2 Information system or information table

An information system or information table can be viewed as a table, consisting of objects
(rows) and attributes (columns). It is used in the representation of data that will be utilized
by Rough Set, where each object has a given amount of attributes (Lin, 1997).
www.intechopen.com
Rough Set Theory – Fundamental Concepts, Principals, Data Extraction, and Applications 37
These objects are described in accordance with the format of the data table, in which rows
are considered objects for analysis and columns as attributes (Wu et al., 2004). Below is
shown an example of an information Table 1.
Attributes
Patient
Headache Vomiting Temperature Viral illness
#1 No Yes High Yes
#2 Yes No High Yes
#3 Yes Yes Very high Yes
#4 No Yes Normal No
#5 Yes No High No
#6 No Yes Very high Yes
Table 1. Example of information table
2.3 Indiscernibility relation

Indiscernibility Relation is a central concept in Rough Set Theory, and is considered as a
relation between two objects or more, where all the values are identical in relation to a
subset of considered attributes. Indiscernibility relation is an equivalence relation, where all
identical objects of set are considered as elementary (Pawlak, 1998)
In Table 1, presented in section 2.2, it can be observed that the set is composed of attributes
that are directly related to the patients' symptoms whether they be headache, vomiting and
temperature. When Table 1 is broken down it can be seen that the set regarding {patient2,
patient3, patient5} is indiscernible in terms of headache attribute. The set concerning
{patient1, patient3, patient4} is indiscernible in terms of vomiting attribute. Patient2 has a
viral illness, whereas patient5 does not, however they are indiscernible with respect to the
attributes headache, vomiting and temperature. Therefore, patient2 and patient5 are the
elements of patients’ set with unconcluded symptoms.
2.4 Approximations
The starting point of rough set theory is the indiscernibility relation, generated by
information concerning objects of interest. The indiscernibility relation is intended to
express the fact that due to the lack of knowledge it is unable to discern some objects
employing the available information Approximations is also other an important concept in
Rough Sets Theory, being associated with the meaning of the approximations topological
operations (Wu et al., 2004). The lower and the upper approximations of a set are interior
and closure operations in a topology generated by the indiscernibility relation. Below is
presented and described the types of approximations that are used in Rough Sets Theory.
a. Lower Approximation (B”)
Lower Approximation is a description of the domain objects that are known with certainty
to belong to the subset of interest.
The Lower Approximation Set of a set X, with regard to R is the set of all of objects, which
certainly can be classified with X regarding R, that is, set B”.
b. Upper Approximation (B*)
Upper Approximation is a description of the objects that possibly belong to the subset of
interest. The Upper Approximation Set of a set X regarding R is the set of all of objects
which can be possibly classified with X regarding R, that is, set B*.
www.intechopen.com
c. Boundary Region (BR)

Boundary Region is description of the objects that of a set X regarding R is the set of all the
objects, which cannot be classified neither as X nor -X regarding R. If the boundary region is
otherwise, if the boundary region is a set X ≠ ∅ (empty) the set X "Rough" is considered. In
a set X =∅ (Empty), then the set is considered "Crisp", that is, exact in relation to R;
Mathematically speaking, let a set X ⊆ U, B be an equivalence relation and a knowledge base

that the boundary region is BR = B* - B”.
K = (U,B). Two subsets can be associated:

1. B-lower: B”= ∪ {Y ∈ U/B : Y ⊆ X}
2. B-upper: B*= ∪ {Y ∈ U/B : Y ∩ X ≠ Ø}
3. POS(B) = B” ⇒ certainly member of X

In the same way, POS(B), BN(B) and NEG(B) are defined below (Pawlak, 1991).
4. NEG(B) = U –B* ⇒ certainly non-member of X

5. BR(B) = B* - B” ⇒ possibly member of X
Figure 1 presents a graphic representation of these regions (Lambert-Torres et al., 1999).
NEG(B)
B* BR(B)
B” POS(B)
Fig. 1. Definition of B-approximation sets and B-regions

d. Quality Approximation
approximation. The coefficient used in measuring the quality is represented by αB(X), where
It is obtained numerically using its own elements, specifically those of lower and upper
X is a set of objects or registrations regarding B. The quality of approximation uses two

coefficients that are presented below:
• Imprecision coefficient αB (X)
Where αB is the quality of approximation of X, being denoted by:
αB(X) = |B”(X)| / |B*(X)| (1)
and the approximation are set ≠ ∅. Therefore, 0 ≤ αB ≤ 1, if αB(X)=1, X it is a definable set

Where |B”(X)| and |B*(X)| it represents the cardinality of approximation lower and upper,
regarding the attributes B, that is, X is crisp set. If αB(X) < 1, X is rough set regarding the
attributes B.
Applying it formulates for Table 1, it has αB(X) = 3/5 for the patients with possibility of they
•
are with viral illness.
Quality Coefficient of upper and lower approximation
www.intechopen.com
- Quality Coefficient of upper approximation αB (B*(X))

It is the percent of all the elements that are classified as belonging to X, being denoted for:
αB(B*(X)) = | B*(X)| / |A| (2)

In the Table 1, αB(B*(X)) =5/6, for the patients that have the possibility of they be with viral
illness.
- Quality Coefficient of lower approximation αB(B”(X))
It is the percentage of all the elements that possibility are classified as belonging to X, and is
denoted as:
αB(B”(X))= |B”(X)| / |A| (3)

In the Table 1, αB(B*(X)) =3/6 = 1/2, for the patients that have viral illness.
Observation: In Quality coefficient upper and lower, presented in number 2, of this section,
|A| represents the cardinality of any given set of objects.
2.5 Decision tables and decision algorithms

A decision table contains two types of attributes designated as the condition attribute and
decision attribute. In Table 1, shown in section 2.2, the attributes of headache, vomiting and
temperature can all be considered as condition attributes, whereas the viral illness attribute
is considered a decision attribute.
Each row of a decision table determines a decision rule, which specifies the decisions
(actions) that must be taken when conditions are indicated by condition attributes are
satisfied, e.g. in Table 1 the condition (Headache, no), (vomiting, yes), (Temperature, high)
determines the decision (Viral illness, yes).
Table 1 shows that both patient2 and patient5 suffer from the same symptoms since the
condition attributes of headache, vomiting and temperature possess identical values;
however, the values of decision attribute differ. These set of rules are known as either
inconsistency, non-determinant or conflicting. These rules are known as consistency,
determinant or non conflicting or simply, a rule.
consistence, which can be denoted by γ(C, D), where C is the condition and D the decision. If
The number of consistency rules, contained in the decision table are known as a factor of
γ(C,D) = 1, the decision table is consistent, but if γ(C,D) ≠ 1 the table of decision is
Given that Table 1, γ(C,D) = 4/6, that is, the Table 1 possesses two inconsistent rules
inconsistent.
(patient2, patient5) and four consistent rules (patient1, patient3, patient4, patient6), inside of
universe of six rules for all the Table 1 (Ziarko & Shan, 1995). The decision rules are
frequently shown as implications in the form of “if... then... “. To proceed is shown one rule
for the implication viral illness:
If
Headache = no and
Vomiting = yes and
Temperature = high
Then
Viral Illness = yes
www.intechopen.com
A set of decision rules is designated as decision algorithms, because for each decision table it
can be associated with the decision algorithm, consisting of all the decision rules that it
occur in the respective decision table. A may be made distinction between decision
algorithm and decision table. A decision table is a data set, whereas a decision algorithm is
a collection of implications, that is, a logical expressions (Pawlak, 1991).
2.6 Dependency of attributes
Intuitively, a set of attributes D depends totally on a set of attributes C, denoted as C ⇒ D, if

In the analysis of data, it is important discover the dependence between attributes.
all values of attributes from D are uniquely determined by values of attributes from C, then
D depends totally on C, if there exists a functional dependency between values of D and C.
For example, in Table 1 there are no total dependencies whatsoever, if in Table 1, the value
of the attribute Temperature for patient p5 was “no” instead of “high”, there would be a
total dependency {Temperature}⇒{viral illness}, because to each value of the attribute
Temperature there would correspond a unique value of the attribute viral illness.
It would also necessitate a more global concept of dependency of attributes, designated as
partial dependency of attributes, in Table 1, the temperature attribute determines some
uniquely values of the attribute viral illness. That is, (temperature, very high) implies (viral
illness, yes), similarly (temperature, normal) implies (viral illness, no), but (temperature,
high) does not imply always (viral illness, yes). Thus the partial dependency means that
only some values of D are determined by values of C. Formally the dependence among the
that D depends on C in degree K (0 ≤ k ≤ 1), denoted ⇒kD and if k = γ(C, D). If K=1, D
attributes can be defined in the following way: If D and C are subsets of A, can be affirmed
depends totally on C, if K < 1, it is said that D depends partially (in a degree K) on C.

The concept of dependent attributes is strongly coupled to the concept of consistency of
decision table (Pawlak, 1991). For example, for dependency attributes {Headache, Vomiting,
Temperature}⇒{viral illness}, it get k=4/6= 2/3, because four out of six patients can be
uniquely classified as having viral illness or not, employing attributes Headache, Vomiting
and Temperature.
It can be interesting to note exactly how patients can be diagnosed using only the attribute
Temperature, that is, the degree of the dependence {Temperature}⇒{viral illness}, diagnose
is obtained k = 3/6 = 1/2, in this case there are only three patients {patient1, patient3,
patient6} out of six, only these tree, can be classified exclusively as having viral illness. In
contrast, with the case of patient4, who cannot be classified as having viral illness, since the
value of the attribute temperature, in this case, is normal. Hence the single attribute
Temperature offers worse classification than the whole set of attributes Headache, Vomiting
and Temperature. It is interesting to observe that neither Headache nor Vomiting can be
used to recognize viral illness, because for both dependencies {Headache}⇒{viral illness}
It can be easily seen that if D depends totally on C then I(C) ⊆ I(D). That means that the
and {Vomiting}⇒{Viral illness} it has k=0.
partition generated by C is finer than the partition generated by D, and that the concept of
dependency presented in the section corresponds to that considered in relational databases.
2.7 Reduction attributes in information system

For many application problems, it is often necessary to maintain a concise form of the
information system, but there exist data that can be removed, without altering the basic
www.intechopen.com
properties and more importantly the consistency of the system (Cerchiari et al, 2006). If is
subtract relative data from the headache and vomiting, the resultant data set is equivalent to
original data in relation to approximation and dependency, as it has the same the
approximation precision and the same dependency degree using the original set of
attributes, however with one fundamental difference, the set of attributes to be considered
will be fewer.
The process of reducing an information system such that the set of attributes of the reduced
information system is independent and no attribute can be eliminated further without losing
subset B ⊆ A preserves the indiscernibility relation RA, then the attributes A - B are
some information from the system, the result is known as reduct. If an attribute from the
dispensable. Reducts are such subsets minimal, i.e., that do not contain any dispensable
attributes. Therefore, the reduction should have the capacity to classify objects, without
altering the form of representing the knowledge (Geng & Zhu, 2006). When the definition
above is applied, the information system presented in the Table 1, B is a subset of A and a
•
belongs to B:
•
a is dispensable in B if I (B) = I (B - {a}); otherwise a is indispensable in B;
•
Set B is independent if all its attributes are indispensable;
Subset B' of B is a reduct of B if B' it is independent and I (B') = I (B); and
A reduct is a set of attributes that preserve the basic characteristics of the original data set;
therefore, the attributes that do not belong to a reduct are superfluous with regard to
classification of elements of the Universe.
3. Rough set with tools for data mining

The great advances in information technology have made it possible to store a great
quantity of data. In the late nineties, the capabilities of both generating and collecting data
were increased rapidly. Millions of databases have been used in business management,
government administration, scientific and engineering data management, as well as many
other applications. It can be noted that the number of such databases keeps growing rapidly
because of the availability of powerful database systems. This explosive growth in data and
databases has generated an urgent need for new techniques and tools that can intelligently
and automatically transform the processed data into useful information and knowledge.
One of the processes used to transform data into knowledge is Knowledge Discovery
Database (KDD), which is divided in three stages (preprocessing, data mining and post
processing) that are shown in the section 3.1.
3.1 Knowledge Discovery in Database - KDD

Knowledge Discovery in Database - KDD is a process, with several stages, no trivial,
interactive and iterative, for the identification of comprehensible patterns, valid, new and
potentially useful starting from great groups of data (Fayyad et al., 1996a). KDD is
characterized as a process composed of several operational stages: preprocessing, data
mining and the post processing. Figure 2 presents the sequence of the stages executed
during the process of KDD.
a. Preprocessing Stage
The preprocessing stage understands the functions related to the reception, the organization
and to the treatment of data, this stage has as its objective the preparation of the data for the
following stage of the data mining.
www.intechopen.com
Users
Preprocessing Data Mining Post Processing
Fig. 2. The KDD process

b. Data Mining Stage
The data mining stage defines the techniques and the algorithms to be used by the problem
in question, as are examples of techniques that can be used in this stage such as neural
network, rough set, genetic algorithms, statistical models and probabilistic. The choice of
technique depends, in many cases, on task type to be developed. In Table 2 is shown a
summary of tasks to be accomplished and some alternative methods that can be useful. It is
important to observe that Table 2 does not drain the universe of methods of data mining
that can be used in each task of KDD and is purely a summary (Fayyad et al., 1996b).
Tasks of KDD Methods of Data Mining

Discovery Associations Basic, Apriori, DHP, Partition, DIC, ASCX-2P
Discovery Generalize of Basic, Apriori, DHP, Partition, DIC, ASCX-2P
Associations
Discovery of Sequences GSP, MSDD, SPADE
Discovery Generalize of Sequences GSP, MSD, SPADE
Classification Neural Network, C4.5, Rough Sets, Genetic
Algorithms, CART, K-NN, Bayes Classifier
Regression Neural Network, Fuzzy Set
Summarization C4.5, Genetics Algorithms
Clustering K-Means, K-Modes, K-prototypes, Fuzzy K-
Means, genetics Algorithms, Neural Network
Forecast of Temporal Series Neural Network, Fuzzy Set
Table 2. Methods of Data Mining that can be applied in the tasks of KDD
During the data mining stage much useful knowledge is gained is respect of the application.
Many authors consider data mining synonymous with KDD, in the context this stage, the
www.intechopen.com
KDD process is often known as Data Mining; in the research it will Data Mining be
indented as KDD (Piatetsky-Shapiro & Matheus, 1995; Mitchell, 1999; Wei, 2003).
Data mining has become an area of research increasing importance, and is also referred to as
knowledge discovery in databases (KDD), consequently this has resulted in a process of non
trivial extraction of implicit, previously unknown and potentially useful information, such
as knowledge rules, constraints, regularities from data in databases
c. Post Processing Stage
In the post processing stage the treatment of knowledge obtained during the data mining
stage. This stage is not always necessary; however, it allows the possibility of validation of
the usefulness of the discovered knowledge.
3.2 Rough set in data mining

Rough set theory constitutes a consistency base for data mining; it offers useful tools for
discovering patterns hidden in data in many aspects. Although in theory rough set deals
with discreet data, rough set is commonly used in conjunction with other techniques
connected to discretization on the dataset. The main feature of rough set data analysis is
both non-invasive and notable ability to handle qualitative data. This fits into most real life
applications nicely.
Rough Set can be used in different phases of the knowledge discovery process, as attribute
selection, attribute extraction, data reduction, decision rule generation and pattern
extraction (Komorowski et al., 1999). Furthermore, recent extensions of rough set theory
with rough mereology have brought new methods of decomposition of large data sets, data
mining in distributed and multi-agent based environments and granular computing
(Polkowski, 2002). It includes mechanisms for defining partial memberships of sets, but does
not introduce additional measures of probabilities or degrees of membership.
The basic idea is that there is some information or data associated with each object in the
universe of discourse. Based on this information, it is possible to tell some of the objects
apart, while others are impossible to distinguish. The latter objects are indiscernible from
each other, and form a set. Each set of indiscernible objects is a knowledge granule, and they
form the building blocks for knowledge about the universe. The rough set community has
been a very active research community since its inception in the eighties, and a large
number of rough set methods for knowledge discovery and data mining have been
developed. The entire knowledge discovery process has been subject to research, and a wide
range of contributions has been made.
Data mining technology provides a new thought for organizing and managing tremendous
data. Rough set theory is one of the important methods for knowledge discovery. This
method can analyze intact data, obtain uncertain knowledge and offer an effective tool by
reasoning.
Rough set has shed light on many research areas, but seldom found its way into real world
application. Data mining with rough set is a multi-phase process consisted of mainly:
discretization; reducts and rules generation on training set; classification on test set.
Rough Set Theory, since it was put forward, has been widely used in Data Mining, and has
important functions in the expression, study and conclusion of uncertain knowledge, it is a
powerful tool, which sets up the intelligent decision system. The main focus is to show how
www.intechopen.com
rough set techniques can be employed as an approach to the problem of data mining and
knowledge extraction.
4. Applications of rough set theory

Rough set theory offers effective methods that are applicable in many branches of Artificial
Intelligence, one of the advantages of rough set theory is that programs implementing its
methods may easily run on parallel computers, but several problems remain to be solved.
Recently, much research has been carried out in Rough Set together with other artificial
Intelligence methods such as Fuzzy Logic, Neural Networks, and Expert Systems and some
significant results have been found. Rough set theory allows characterization of a set of
objects in terms of attribute values; finding dependencies total or partial between attributes;
reduction of superfluous attributes; finding significance attributes and decision rule
generation.
The applications of Rough Set have resolved complex problems; and therefore have been
attractive to researchers in recent years, already it has been applied successfully in a number
of challenging fields such the a soft computing method.
This section provides a brief overview of some of the many applications of rough set. There
are several properties of rough sets that make the theory an obvious choice for use in
dealing with real problems:
a. Pattern Recognition
Pattern Recognition using Rough Set is one such successful application field, but in 2001 A.
Mrozek and K. Cyran (2001) proposed a hybrid method of automatic diffraction pattern
recognition based on Rough Set Theory and Neural Network. In this new method, the rough
set is used to define the objective function and stochastic evolutionary algorithm for space
search of a feature extractor, and neural networks are employed to model the uncertain
systems. The features obtain end by optimized sampling of diffraction patterns are input to
a semantic classifier and the pattern recognition algorithm is performed with optimized and
standard computer-generated holograms.
b. Emergency room diagnostic medical
A common and diagnostically challenging problem facing emergency department personnel
in hospitals is that of acute abdominal pain in children. There are many potential causes for
this pain, most usually non-serious. However, the pain may be an indicator that a patient
has a serious is illness, requiring immediate treatment and possibly surgery. Experienced
doctors will use a variety of relevant historical information and physical observations to
assess children. Such attributes occur frequently in recognizable patterns, allowing a quick
and efficient diagnosis. Inexperienced doctors, on the other hand, may lack the knowledge
and information to be able to recognize these patterns. The rough set based clinical decision
model is used to assist such inexperienced doctors. In this research, rough sets are used to
support diagnosis by distinguishing between three disposition categories: discharge,
observation/further investigation, and consultation. Preliminary results show that the
system gives accuracy comparable to doctors, though it is dependent on a suitably high data
quality (Rubin et al., 1996)
c. Acoustical analysis
Rough Set was applied for the assessment of concert hall acoustics. Rough set algorithms are
applied to the decision table containing subjectively quantified parameters and the results of
www.intechopen.com
overall subjective preference of acoustical objects described by the parameters. Fuzzy

membership functions map the test results to approximate the tested parameter distribution,
which is determined on the basis of the separate subjective test of individual parameter
underlying overall preference. A prototype system based on the rough set theory is used to
induce generalized rules that describe the relationship between acoustical parameters of
concert halls and sound processing algorithms (Kotek, 1999)
d. Power system security analysis
Rough Set is a systematic approach used to help knowledge engineers during the extraction
process of facts and rules of a set of examples for power system operation problems. This
approach describes the reduction the number of examples, offering a more compact set of
examples to the user (Lambert-Torres et al., 1999).
e. Spatial and meteorological pattern classification
Some categories of sunspot groups are associated with solar flares. Observatories around the
world track all visible sunspots in an effort to detect flares early, the sunspot recognition
and classification are currently manual and labor intensive processes which could be
automated if successfully learned by a machine. The approach employs a hierarchical rough
set based learning method for sunspot classification. It attempts to learn the modified Zurich
classification scheme through rough set-based decision tree induction. The resulting system
has been evaluated on sunspots extracted from satellite images, with promising results
(Nguyen et al., 2005).
A new application of rough set theory for classifying meteorological radar data has been
introduced. Volumetric radar data is used to detect storm events responsible for severe
weather. Classifying storm cells is a difficult problem as they exhibit a complex evolution
throughout their lifespan. Also, the high dimensionality and imprecision of the data can be
prohibitive. Rough set approach is employed to classify a number of meteorological storm
events (Shen & Jensen, 2007).
f. Intelligent control systems
An important application field of rough set theory is that of intelligent control systems
especially when incorporated with fuzzy theory (Xie et al., 2004).
g. Measure the quality of a single subset
Ant Colony System algorithm and Rough Set Theory proposed a hybrid approach to feature
selection, in Rough Set Theory offers a heuristic function in order to measure the quality of a
single subset. It has studied the influence of the setting of the parameters for this problem, in
particular for finding a reduct. Experimental results show this hybrid approach is a potential
method for features selection (He et al., 2007).
There are infinite possibilities in the development of methods based on Rough Set Theory
such as nonstandard analysis, nonparametric statistics and qualitative.
5. Case – rough set with tools in dengue diagnosis

In this section, several patients data set is shown with possible dengue symptoms. Through
data are analysis is accomplished, using a Rough Set approach for the elimination of
redundant data and the development of a set of rules that it can aid the doctor in the
elaboration of the diagnosis. Below the Table 3 is shown with the patients data set and
respective symptoms, and the data are of the discreet type.
www.intechopen.com
5.1 Information system or information table

Decision
Conditional Attributes
Attribute
Patient
muscular_pain_articu
blotched_red_skin Temperature Dengue
lations
P1 No No Normal No
P2 No No High No
P3 No No Very High Yes
P4 No Yes High Yes
P5 No Yes Very High Yes
P6 Yes Yes High Yes
P7 Yes Yes Very High Yes
P8 No No High No
P9 Yes No Very High Yes
P10 Yes No High No
P11 Yes No Very High No
P12 No Yes Normal No
P13 No Yes High Yes
P15 Yes Yes Normal No
P16 Yes No Normal No
P17 Yes No High No
Table 3. Patients with respective symptoms
Where, B are all of the objects or registrations of the system, given set B={P1, P2, P3, P4, P5,
P6, P7, P8, P9, P10, P11, P12, P13, P14, P15, P16, P17, P18, P19, P20} the set conditional
attributes is represented by C={blotched_red_skin, muscular_pain_articulations,
Temperature} and the set D represented the decision attribute, where D={dengue}. The set A
or Table 3, can be shown in relation to the function of nominal values of considered
attributes, in the Table 4:
Attributes Nominal Values
blotched_red_skin, Yes, No
Conditional
muscular_pain_articulations Yes, No
Attributes
Temperature Normal, High, Very High
Decision Attributes Dengue Yes, No
Table 4. Nominal Values of Attributes
5.2 Indiscernibility relation

Indiscernibility Relation is the relation between two objects or more, where all the values are
identical in relation to a subset of considered attributes. In Table 3, presented in section 5.1,
it can be observed that the set is composed of attributes that are directly related to patients'
www.intechopen.com
symptoms, where C={blotched_red_skin, muscular_pain_articulations, temperature}, the

indiscernibility relation is given to INDA(C). When Table 3 is broken down it can be seen
that indiscernibility relation is given in relationship to conditional attributes:
- The blotched_red_skin attribute generates two indiscernibility elementary sets:
INDA({blotched_red_skin})={{P1,P3,P4,P5,P8,P12,P13,P14,P20},{P6,P7,P9,P10,P11,P15,P
16, P17,P18,P19}}.
Decision
Attribute
Patient
lations
P1 No No Normal No
P13 No Yes High Yes
P2 No Yes High No
P4 No Yes High Yes
P8 No No High No
P10 Yes No High No
P17 Yes No High No
P6 Yes Yes High Yes
Table 5. Table 3 organize in relations blotched_red_skin attribute
- The muscular_pain_articulations attribute generates two indiscernibility elementary
sets: INDA({muscular_pain_articulations})={{P1, P2, P3, P8, P9, P10, P11, P16, P17, P19},
{P4, P5, P6, P7, P12, P13, P14, P15, P18, P20}}.
Decision
Attribute
Patient
lations
P1 No No Normal No
P2 No No High No
P8 No No High No
P10 Yes No High No
www.intechopen.com

P17 Yes No High no
P4 No Yes High Yes
P6 Yes Yes High Yes
P13 No Yes High Yes
Table 6. Table 3 organize in relation muscular_pain_articulations attribute
- The temperature attribute generates three indiscernibility elementary sets:
INDA({temperature}) = {{P2, P4, P6, P8, P10, P13, P17} , {P3, P5, P7, P9, P11, P18} , {P1,
P12, P14, P15, P16, P19, P20}}.
Decision
Attribute
Patient
lations
P13 No Yes High Yes
P2 No Yes High No
P4 No Yes High Yes
P8 No No High No
P10 Yes No High No
P17 Yes No High No
P6 Yes Yes High Yes
P1 No No Normal No
Table 7. Table 3 organized in relation temperature attribute
www.intechopen.com
5.3 Approximation
The lower and the upper approximations of a set are interior and closure operations in a
topology generated by a indiscernibility relation. Below is presented and described the types
of approximations are followed using in Rough Set Theory; the approximations concepts are
applied in the Table 3, shown to proceed:
Decision
Attribute
Patient
lations
P1 No No Normal No
P2 No No High No
P8 No No High No
P10 Yes No High No
P17 Yes No High No
P4 No Yes High Yes
P6 Yes Yes High Yes
P13 No Yes High Yes
Table 8. Table 3 organized in relation decision attribute
a. Lower Approximation set B”
- Lower Approximation set (B”) of the patients that are definitely have dengue are
identified as B” = {P3,P4,P5,P6,P7,P13,P18}
- Lower Approximation set (B”) of patients that certain have not dengue are identified as
B” = {P1 ,P2 ,P8 ,P10 ,P12, P14, P15, P16, P17, P19,P20}
b. Upper Approximation set B*
- Upper Approximation set (B*) of the patients that possibly have dengue are identified
as B* = {P3,P4,P5,P6,P7, P9, P13,P18}
- Upper Approximation set (B*) of the patients that possibly have not dengue are
identified as B* = {P1, P2, P8, P10, P11, P12, P14, P15, P16, P17, P19, P20}
c. Boundary Region (BR)
- Boundary Region (B*) of the patients that not have dengue are identified as:
BR = {P1,P2,P8,P10,P11,P12,P14,P15,P16,P17,P19,P20} -
{P1,P2,P8,P10,P12,P14,P15,P16,P17, P19,P20} = {P11};
www.intechopen.com
- Boundary Region (B*), the set of the patients that have dengue are identified as: BR =
{P3,P4,P5,P6,P7, P9, P13,P18} - {P3,P4,P5,P6,P7,P13,P18} = {P9}
Observation: Boundary Region (BR), the set constituted by elements P9 and P11, which
cannot be classified, since they possess the same characteristics, but with differing
conclusions differ in the decision attribute.
5.4 Quality of approximations

The two coefficients of quality of approximation are:
•
- Imprecision coefficient, using Eq. (1)):
for the patients with possibility of they are with dengue αB(X) = =7/8;
• for the patients with possibility of they are not with dengue αB(X) = 8/12.
•
- Quality Coefficient of upper and lower approximation, using Eq. 2 and 3:
αB(B*(X)) =8/20, for the patients that have the possibility of they be with dengue;
• αB(B*(X)) =11/20, for the patients that not have the possibility of they be with
•
dengue;
αB(B”(X)) =7/20, for the patients that have dengue;
• αB(B”(X)) =8/20, for the patients that not have dengue.
Observations:
1. Patient with dengue: αB(B”(X))=7/20, that is, 35% of patients certainly with
dengue.
2. Patient that don't have dengue: αB(B”(X)) = 11/20, that is, approximately 55% of
patients certainly don't have dengue.
3. 10% of patients (P9 and P11) cannot be classified neither with dengue nor without
dengue, since the characteristics of all attributes are the same, with only the
decision attribute (dengue) not being identical and generates an inconclusive
diagnosis for dengue.
6. Data reduction in information system

The form in which data is presented within an information system must guarantee that the
redundancy is avoided as it implicates the minimization of the complexly computational in
relation to the creation of rules to aid the extraction knowledge. However, when the
information system possesses redundancy situations, it is necessary to treat it One of the
ways of accomplishing this is to use the concept of reduct, without altering the
indiscernibility relations.
A reduct is a set of necessary minimum data, since the original proprieties of the system or
information table are maintained. Therefore, the reduct must have the capacity to classify
objects, without altering the form of representing the knowledge.
The process of reduction of information is presented below in Table 3, it can be observed
that the data is of a discreet type.
a. Verification inconclusive data
Step 1 – Analysis of data contained in Table 3 shows that possess information inconclusive,
being that the values of conditional attributes same and the value of decision attribute is
different.
Conclusion of Step 1: The symptoms of patient P9 and patient P11 are both inconclusive,
since they possess equal values of conditions attributes together with a value of decision
www.intechopen.com
attribute that is different. Therefore, the data of patient P9 and patient P11 will be excluded
from Table 3.
b. Verification of equivalent information
Step 2 – Analysis of data contained in Table 3 shows that it possesses equivalent
information.
P2 No No High No
P8 No No High No
P4 No Yes High Yes

P13 No Yes High Yes

P10 Yes No High No

P17 Yes No High No


Conclusion of Step 2 – The Table 3 has it reduced data presented in a revised version in
Table 9 shown below:
Decision
Attribute
Patient
lations
P1 No No Normal No
P2 No No High No
P4 No Yes High Yes
P6 Yes Yes High Yes
P8 No No High No
P10 Yes No High No
Table 9. Reduct of Information of Table 3
Step 3 – Analysis of each condition attributes with the attributes set.
www.intechopen.com
Conditional Attributes Decision Attribute

Patient
blotched_red_skin Dengue
P1 No No
P2 No No
P3 No Yes
P4 No Yes
P5 No Yes
P6 Yes Yes
P7 Yes Yes
P8 No No
P10 Yes No
P12 No No
P15 Yes No
P16 Yes No
P19 Yes No
Table 10. Analysis of Attribute blotched_red_skin in Table 9
Conditional Attributes Decision Attribute

Patient
muscular_pain_articulations Dengue
P1 No No
P2 No No
P3 No Yes
P4 Yes Yes
P5 Yes Yes
P6 Yes Yes
P7 Yes Yes
P8 No No
P10 No No
P12 Yes No
P15 Yes No
P16 No No
P19 No No
Table 11. Analysis of Attribute muscular_pain_articulations in Table 9
Conditional Attribute Decision Attribute

Patient
Temperature Dengue
P1 Normal No
P2 High No
P3 Very High Yes
P4 High Yes
P5 Very High Yes
P6 High Yes
www.intechopen.com
P7 Very High Yes

P8 High No
P10 High No
P12 Normal No
P15 Normal No
P16 Normal No
P19 Normal No
Table 12. Analysis of Attribute Temperature in Table 9
Conclusion of analysis: In this analysis, no data was excluded.
c. Given analysis of condition attributes in Table 9, it can be observed that the same data
exists in proceeding tables.
- Analysis of attributes blotched_red_skin and muscular_pain_articulations in Table 9.
Decision
Attribute
Patient
muscular_pain_articula
tions
P1 No No No
P2 No No No
P3 No No Yes
P4 No Yes Yes
P5 No Yes Yes
P12 No Yes No
P6 Yes Yes Yes
P7 Yes Yes Yes
P8 No No No
P16 Yes No No
P19 Yes No No
P10 Yes No No
P15 Yes Yes No
Table 13. Analysis of Attributes blotched_red_skin and muscular_pain_articulations in
Table 9
Result of analysis
Decision
Attribute
Patient
tions
P1 No No No
P3 No No Yes
P4 No Yes Yes
P6 Yes Yes Yes
P10 Yes No No
P15 Yes Yes No
Table 14. Result of Analysis of Attributes blotched_red_skin and muscular_pain_articulations
in Table 9
www.intechopen.com
- Analysis of Attributes attributes blotched_red_skin and temperature.

Decision
Patient Attribute
P1 No Normal No
P12 No Normal No
P2 No High No
P8 No High No
P3 No Very High Yes
P5 No Very High Yes
P7 Yes Very High Yes
P4 No High Yes
P6 Yes High Yes
P10 Yes High No
P15 Yes Normal No
P16 Yes Normal No
P19 Yes Normal No
Table 15. Analysis of Attributes blotched_red_skin and temperature in Table 9
Result of analysis
Decision
Patient Attribute
P1 No Normal No
P2 No High No
P3 No Very High Yes
P4 No High Yes
P6 Yes High Yes
P10 Yes High No
P15 Yes Normal No
Table 16. Result of it Analysis of Attributes blotched_red_skin and temperature in Table 9
- Analysis of attributes muscular_pain_articulations and temperature in Table 9.
Decision
Attribute
Patient
muscular_pain_articulati
Temperature Dengue
ons
P1 No Normal No
P16 No Normal No
P19 No Normal No
P2 No High No
P8 No High No
P10 No High No
www.intechopen.com
P3 No Very High Yes

P4 Yes High Yes
P6 Yes High Yes
P12 Yes Normal No
P15 Yes Normal No
Table 17. Analysis of Attributes muscular_pain_articulations and temperature in Table 9
Result of analysis
Decision
Attribute
Patient
muscular_pain_articulati
Temperature Dengue
ons
P1 No Normal No
P2 No High No
P3 No Very High Yes
P4 Yes High Yes
P12 Yes Normal No
Table 18. Result of it analysis of Attributes muscular_pain_articulations and temperature in
Table 9
Step 4 – Verification of equivalent (intersection) data in the Tables 14, 16 and 18 correspond
where data is the element of reduct information in relation to Table 9.
Decision
Attribute
Patient
tions
P1 No No Normal No
P4 No Yes High Yes
Table 19. Table with result of information reduct of Table 9
6. Decision rules
With the information reduct shown above, it can be generated the necessary decision rules
for aid to the dengue diagnosis. The rules are presented to proceed:
Rule-1
R1: If patient
blotched_red_skin = No and
muscular_pain_articulations = No and
temperature = Normal
Then dengue = No.
www.intechopen.com
Rule-2
R2: If patient
muscular_pain_articulations = No and
temperature = Very High
Then dengue = Yes.
Rule-3
R3: If patient
muscular_pain_articulations = Yes and
temperature = High
Then dengue = Yes.
7. Conclusion
This study, it has discussed the Rough set theory, was proposed in 1982 by Z. Pawlak, as an
approach to knowledge discovery from incomplete, vagueness and uncertain data. The
rough set approach to processing of incomplete data is based on the lower and the upper
approximation, and the theory is defined as a pair of two crisp sets corresponding to
approximations.
The main advantage of rough set theory in data analysis is that it does not need any
preliminary or additional information concerning data, such as basic probability assignment
in Dempster-Shafer theory, grade of membership or the value of possibility in fuzzy set
theory. The Rough Set approach to analysis has many important advantages such as
(Pawlak, 1997): Finding hidden patterns in data; Finds minimal sets of data (data reduction);
Evaluates significance of data; Generates sets of decision rules from data; Facilitates the
interpretation of obtained result
Different problems can be addressed though Rough Set Theory, however during the last few
years this formalism has been approached as a tool used with different areas of research.
There has been research concerning be relationship between Rough Set Theory and the
Dempster-Shafer Theory and between rough sets and fuzzy sets. Rough set theory has also
provided the necessary formalism and ideas for the development of some propositional
machine learning systems.
Rough set has also been used for knowledge representation; data mining; dealing with
imperfect data; reducing knowledge representation and for analyzing attribute
dependencies.
Rough set Theory has found many applications such as power system security analysis,
medical data, finance, voice recognition and image processing; and one of the research areas
that has successfully used Rough Set is the knowledge discovery or Data Mining in
database.
8. References
Cerchiari, S.C.; Teurya, A.; Pinto, J.O.P.; Lambert-Torres, G.; Sauer, L. & Zorzate, E.H. (2006).
Data Mining in Distribution Consumer Database using Rough Sets and Self-
Organizing Maps, Proceedings of the 2006 IEEE Power Systems Conference and
www.intechopen.com
Exposition, pp. 38-43, ISBN 1-4244-0177-1, Atlanta–USA, Oct. 29–Nov. 1, 2006, IEEE
Press, New Jersey-USA.
Fayyad, U.; Piatetsky-Shapiro, G. & Smyth, P. (1996a). From Data Mining to Knowledge
Discovery: An Overview, In : Advances in Knowledge Discovery & Data Mining,
Fayyad, U.; Piatetsky-Shapiro, G.; Smyth, P. & Uthurusamy, R. (Ed.), pp. 1–34.
AAAI Press, ISBN 978-0-262-56097-9, Menlo Park-USA.
Fayyad, U., G. Piatetsky-Shapiro, & P. Smyth. (1996b). Knowledge Discovery and Data
Mining: Towards an Unifying Framework, The Proceedings of the Second International
Conference on Knowledge Discovery and Data Mining, pp. 82–88, ISBN 978-1-57735-
004-0, Portland–USA, Aug. 2–4, 1996, AAAI Press, Menlo Park-USA.
Geng, Z. & Qunxiong, Z. (2006). A New Rough Set-Based Heuristic Algorithm for Attribute
Reduct, Proceedings of the 6th World Congress on Intelligent Control and Automation,
pp. 3085-3089, ISBN 1-4244-0332-4, Dalian-China, Jun. 21-23, 2006, Wuhan
University of Technology Press, Wuhan-China.
He, Y.; Chen, D.; Zhao, W. (2007). Integration Method of Ant Colony Algorithm and Rough
Set Theory for Simultaneous Real Value Attribute Discretization and Attribute
Reduction, In: Swarm Intelligence: Focus on Ant and Particle Swarm Optimization,
Chan, F.T. S. & Tiwari, M.K. (Ed.), pp. 15–36, I-TECH Education and Publishing.
ISBN 978-3-902613-09-7, Budapest-Hungary.
Komorowski, J.; Pawlak, Z.; Polkowski, L. & Skowron, A. (1999). Rough Sets Perspective on
Data and Knowledge, In: The Handbook of Data Mining and Knowledge Discovery,
Klosgrn, W. & Zylkon, J. (Ed.), pp. 134–149, Oxford University Press, ISBN 0-19-
511831-6, New York-USA.
Kostek, B. (1999). Assessment of Concert Hall Acoustics using Rough Set and Fuzzy Set
Approach, In: Rough Fuzzy Hybridization: A New Trend in Decision-Making, Pal, S. &
Skowron, A. (Ed.), pp. 381-396, Springer-Verlag Co., ISBN 981-4021-00-8, Secaucus-
USA.
Lambert-Torres, G.; Rossi, R.; Jardini, J.A.; Alves da Silva, A.P. & Quintana, V.H. (1999).
Power System Security Analysis based on Rough Classification, In: Rough Fuzzy
Hybridization: A New Trend in Decision-Making, Pal, S. & Skowron, A. (Ed.), pp. 263-
300, Springer-Verlag Co., ISBN 981-4021-00-8, Secaucus-USA.
Lin, T. Y. (1997). An Overview of Rough Set Theory from the Point of View of Relational
Databases, Bulletin of International Rough Set Society, Vol. 1, No. 1, Mar. 1997, pp. 30-
34, ISSN 1346-0013.
Mitchell, T. M. (1999). Machine learning and data mining, Communications of the ACM, Vol.
42, No. 11, Nov. 1999, pp. 30-36, ISSN 0001-0782.
Mrozek, A. & Cyran, K. (2001). Rough Set in Hybrid Methods for Pattern Recognition,
International Journal of Intelligence Systems, Vol. 16. No. 2, Feb. 2001, pp.149-168,
ISSN 0884-8173.
Nguyen, S.H.; Nguyen, T.T. & Nguyen, H.S. (2005). Rough Set Approach to Sunspot
Classification Problem, Proceedings of the 2005 International Conference on Rough Sets,
Fuzzy Sets, Data Mining and Granular Computing - Lecture Notes in Artificial
Intelligence 3642, pp. 263–272, ISBN 978-3-540-28653-0, Regina-Canada, Aug. 31-
Sept. 3, 2005, Springer, Secaucus-USA.
Piatetsky-Shapiro, G. & Matheus, C. J. (1995). The Interestingness of Deviations, Proceedings
of the International Conference on Knowledge Discovery and Data Mining, pp. 23–36,
www.intechopen.com
ISBN 978-0-929280-82-0, Montreal-Canada, Aug. 1995, AAAI Press, Menlo Park-

USA.
Pawlak, Z. (1982). Rough Sets, International Journal of Information and Computer Sciences, Vol.
11, pp. 341- 356, 1982.
Pawlak, Z. (1991). Rough Sets: Theoretical Aspects of Reasoning about Data., Kluwer Academic
Publishers, ISBN 0-79231472, Norwell-USA.
Pawlak, Z.; Grzymala-Busse, J. ; Slowinski, R. & Ziarko, W. (1995) Rough Set,
Communications of the ACM, Vol. 38, No. 11, Nov. 1995, pp. 88-95, ISSN 0001-0782.
Pawlak, Z. (1997). Vagueness - a Rough Set View, In: Lecture Notes in Computer Science- 1261,
Mycielski, J; Rozenberg, G. & Salomaa, A. (Ed.), pp. 106-117, Springer, ISBN 3-540-
63246-8, Secaucus-USA.
Pawlak, Z. (1998). Granularity of Knowledge, Indiscernibility and Rough Sets, The 1998 IEEE
International Conference on Fuzzy Systems Proceedings - IEEE World Congress on
Computational Intelligence, pp. 106-110, ISBN 0-7803-4863-X, May 4-9, 1998,
Anchorage-USA, IEEE Press, New Jersey-USA.
Rubin, S.; Michalowski, W. & Slowinski, R. (1996). Developing an Emergency Room
Diagnostic Check List using Rough Sets: A Case Study of Appendicitis, In:
Simulation in the Medical Sciences, Anderson, J. & Katzper, M. (Ed.), pp. 19-24, The
Society for Computer Simulation Press, San Diego-USA.
Shen, Q. & Jensen, R. (2007). Rough Sets, Their Extensions and Applications, International
Journal of Automation and Computing, Vol. 4, No. 3, Jul. 2007, pp. 217-228, ISSN 1476-
8186.
Stoll, R.R. (1979). Set Theory and Logic, Dover Publications, ISBN 0-486-63829-4, Mineola-
USA.
Xie, G.; Wang, F. & Xie, K. (2004). RST-Based System Design of Hybrid Intelligent Control,
Proceedings of the 2004 IEEE International Conference on Systems, Man and Cybernetics,
pp. 5800-5805, ISBN 0-7803-8566-7, The Hague-The Netherlands, Oct. 10-13, 2004,
IEEE Press, New Jersey-USA.
Zadeh, L. A. (1965). Fuzzy Sets, Information and Control, No. 8, pp. 338-353.
Ziarko, W. & Shan, N. (1995). Discovering Attribute Relationships, Dependencies and Rules
by using Rough Sets, Proceedings of the 28th Annual Hawaii International Conference on
System Sciences, pp. 293-299, ISBN 0-8186-6930-6, Wailea-USA, Jan. 3-6, 1995, IEEE
Press, New Jersey-USA.
Wei, J. M. (2003). Rough Set based Approach to Selection of Node, International Journal of
Computational Cognition, Vol. 1, No. 2, pp. 25–40, ISSN 1542-8060.
Wu, C.; Yue, Y.; Li, M. & Adjei, O. (2004). The Rough Set Theory and Applications,
Engineering Computations, Vol. 21, No. 5, pp.488-511, ISSN 0264-4401.
www.intechopen.com
Data Mining and Knowledge Discovery in Real Life Applications
Edited by Julio Ponce and Adem Karahoca
ISBN 978-3-902613-53-0
Hard cover, 436 pages
Publisher I-Tech Education and Publishing
Published online 01, January, 2009
Published in print edition January, 2009
This book presents four different ways of theoretical and practical advances and applications of data mining in
different promising areas like Industrialist, Biological, and Social. Twenty six chapters cover different special
topics with proposed novel ideas. Each chapter gives an overview of the subjects and some of the chapters
have cases with offered data mining solutions. We hope that this book will be a useful aid in showing a right
way for the students, researchers and practitioners in their studies.
How to reference
In order to correctly reference this scholarly work, feel free to copy and paste the following:
Silvia Rissino and Germano Lambert-Torres (2009). Rough Set Theory — Fundamental Concepts, Principals,
Data Extraction, and Applications, Data Mining and Knowledge Discovery in Real Life Applications, Julio Ponce
and Adem Karahoca (Ed.), ISBN: 978-3-902613-53-0, InTech, Available from:
http://www.intechopen.com/books/data_mining_and_knowledge_discovery_in_real_life_applications/rough_set
_theory_-_fundamental_concepts__principals__data_extraction__and_applications
InTech Europe InTech China

University Campus STeP Ri Unit 405, Office Block, Hotel Equatorial Shanghai
Slavka Krautzeka 83/A No.65, Yan An Road (West), Shanghai, 200040, China
51000 Rijeka, Croatia
Phone: +385 (51) 770 447 Phone: +86-21-62489820
Fax: +385 (51) 686 166 Fax: +86-21-62489821
www.intechopen.com
© 2009 The Author(s). Licensee IntechOpen. This chapter is distributed
under the terms of the Creative Commons Attribution-NonCommercial-
ShareAlike-3.0 License, which permits use, distribution and reproduction for
non-commercial purposes, provided the original is properly cited and
derivative works building on this content are distributed under the same
license.

Rough Set Theory

Uploaded by

Copyright:

Available Formats

Rough Set Theory

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Rough Set Theory

Uploaded by

Copyright:

Available Formats

3

Rough Set Theory – Fundamental Concepts,

evaluation of quality and availability of information with respect to consistency and

2.2 Information system or information table

2.3 Indiscernibility relation

c. Boundary Region (BR)

Mathematically speaking, let a set X ⊆ U, B be an equivalence relation and a knowledge base

K = (U,B). Two subsets can be associated:

3. POS(B) = B” ⇒ certainly member of X

4. NEG(B) = U –B* ⇒ certainly non-member of X

Fig. 1. Definition of B-approximation sets and B-regions

X is a set of objects or registrations regarding B. The quality of approximation uses two

αB(X) = |B”(X)| / |B*(X)| (1)

and the approximation are set ≠ ∅. Therefore, 0 ≤ αB ≤ 1, if αB(X)=1, X it is a definable set

- Quality Coefficient of upper approximation αB (B*(X))

αB(B*(X)) = | B*(X)| / |A| (2)

αB(B”(X))= |B”(X)| / |A| (3)

2.5 Decision tables and decision algorithms

2.6 Dependency of attributes

Intuitively, a set of attributes D depends totally on a set of attributes C, denoted as C ⇒ D, if

depends totally on C, if K < 1, it is said that D depends partially (in a degree K) on C.

2.7 Reduction attributes in information system

3. Rough set with tools for data mining

3.1 Knowledge Discovery in Database - KDD

Preprocessing Data Mining Post Processing

Fig. 2. The KDD process

Tasks of KDD Methods of Data Mining

3.2 Rough set in data mining

4. Applications of rough set theory

overall subjective preference of acoustical objects described by the parameters. Fuzzy

5. Case – rough set with tools in dengue diagnosis

5.1 Information system or information table

5.2 Indiscernibility relation

symptoms, where C={blotched_red_skin, muscular_pain_articulations, temperature}, the

P11 Yes No Very High No

5.4 Quality of approximations

6. Data reduction in information system

P4 No Yes High Yes

P7 Yes Yes Very High Yes

P10 Yes No High No

P12 No Yes Normal No

P16 Yes No Normal No

Conditional Attributes Decision Attribute

Conditional Attributes Decision Attribute

Conditional Attribute Decision Attribute

P7 Very High Yes

- Analysis of Attributes attributes blotched_red_skin and temperature.

P3 No Very High Yes

ISBN 978-0-929280-82-0, Montreal-Canada, Aug. 1995, AAAI Press, Menlo Park-

InTech Europe InTech China

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

αB(B(X)) = | B(X)| / |A| (2)