11.using rule based classifiers for the predictive analysis of breast cancer recurrence

Journal of Information Engineering and Applications www.iiste.org
ISSN 2224-5782 (print) ISSN 2225-0506 (online)
Vol 2, No.2, 2012

Using Rule Based Classifiers for the Predictive Analysis of
Breast Cancer Recurrence
Srinivas Murti* , Mahantappa
Department of Computer Science and Engineering,PG Extension Center, Vishweshwarayya
Technological University , Bagalkot, India
* E-mail of the corresponding author: s.sri108@gmail.com

Abstract
The Aim of this work is to assess the Effectiveness of Rule Based Classifiers to help an Oncology Doctor
for prediction of Breast Cancer Recurrence ,286 Cancer patient data ,obtained from UCI Machine learning
Repository ,are used to determine Recurrence Events for New patients .This dataset is processed with
WEKA Data Mining Tool ,by applying Rule Based Classifiers(RIPPER,DT,DTNB) and Rule Set is
generated .Further from Experimental Results, it has been found that DTNB is providing improved
Accuracy compared to other two Classifiers .Based on the patients’ characteristics and the Rule set
generated by DTNB ,New patients may be labeled as developing or not ‘Recurrence Events ’,thus
supporting an Oncology Doctor in making Decisions about disease in a shorter time.
Keywords: Medical Data Mining, KDD, Breast Cancer Recurrence, Classification, Association, WEKA,
Rule Based Classifiers
1. Introduction
Breast Cancer is the leading cause of deaths in Women .One in Nine women is expected to develop Breast
Cancer .Breast cancer can recur at any time, But most of the Breast Cancer recurrences occur in first three
to five years of initial treatment .Breast cancer can come back as a local recurrence (in the treated Breast or
mastectomy scar) or as a distant recurrence somewhere else in the body .
1.1 Problem Definition and our Proposed Technique
The nature of the Medical Data is Noisy, In complete and Uncertain. Too much of the Medical Data are
now collected due to the computerization. Too many Disease markers (attributes) are also available in
decision making. The Relationships and patterns within this Medical Data could provide New Medical
Knowledge and enhance our Knowledge of disease progression and management .Evaluation of stored
medical data using the tools like WEKA (Waikato Environment for Knowledge Analysis) may lead to
discovery of trends and patterns. Techniques are needed to search large quantities of this Medical Data for
these Patterns and Relationships. The major challenge facing Health Industry is the Quality of Service at
affordable costs. A Quality of Service means diagnosing the patients correctly and treating them correctly.
Poor Clinical Decisions can lead to disastrous results which is unacceptable. In this Context Medical Data
Mining came into existence.
Data Mining also referred to as -Knowledge Discovery in Databases or KDD, is the search for
Relationships and Global Patterns that exist in the Large Data Bases but are “hidden” among vast amounts
of data . Classification and Association are two mechanisms to represent Extracted Information. Even most
technologically advanced Hospitals in India have no such software that predicts a disease through KDD
process or through a Data Mining Technique. Our Work attempts to predict efficiently the Breast Cancer
Recurrence (as a particular disease) using Classification Data Mining Technique, Thus helping an
Oncologist to diagnose and treat the disease in a short time and over come the problems stated above that
Persists in the Medical Data and Health Industry.
The rest of the paper is organized as follows. The concept of Rule Based Classifiers and some of the
advanced Rule Based Classifiers - discussed in Section 2; and in Section 3- Methodology for our proposed

12

Vol 2, No.2, 2012

work has been detailed; Section 4 outlines the Results and Section 5 illustrates Conclusions and future
work.
1.2 Literature Survey
Up to Now, Several studies have been reported that have focused on Medical Diagnosis .These studies have
applied different approaches to the given problem and have achieved higher prediction accuracies ranging
from 62% or higher, Using the dataset from UCI Machine Learning Repository .
Belciug,S. ; Gorunescu ,F. ; Salem ,A,-B.; Gorunsecu ,M.; have accessed the effectiveness of three
different Clustering algorithms used to detect Breast Cancer recurrent events and Experimental Results
showed that Best Performance was obtained by Cluster Network followed by SOM and K –means. Based
on the Patients segmentation regarding the occurrence of Recurrent Events, New Patients were labeled
according to their Medical Characteristics as developing or not recurrent events.
Machine learning approaches have focused on models (e.g., neural nets, Bayesian nets, hyperplanes) that
are unfamiliar to most non-analyst users. Although data mining models in the form of if- then rules [Apte
and Hong, 1996; Holte, 1993; Michal ski et al.,1986], decision trees [Quinlan, 1993] and association rules
[Agrawal et al., 1996; Han and Fu, 1995; Liu et al.,1998] are considered to be easy to understand, problems
arise when the size of trees or the number of rules become very large. Many successful applications that
employ association rule mining algorithms tend to produce very large numbers of rules. Recent work on
identifying “interesting” rules and on visualization techniques has been done to improve the
comprehensibility of the models [Bayardo and Agrawal, 1999; Liu et al.,1999]. Decision trees are often
viewed as one of the most comprehensible models. To their surprise, Kohavi and Sommerfield [Kohavi and
Sommerfield, 1998] found that it took longer than expected to explain the meaning of decision tree models
to their clients. Rule Based Classifiers like Decision Table, on the other hand ,are easy to interpret and
explain because of widespread familiarity with the tabular representations in spreadsheets and relational
databases.However, relatively little work has been done on algorithms that use decision tables as
representations of hypotheses induced from data sets,perhaps because complex models tend to be more
accurate [Kohavi and Sommerfield, 1998]. Users however, would often be willing to sacrifice some
accuracy for a more easily interpretable model, and it would be useful to know how well Rule Based
Classifiers like decision tables, DTNB (Decision Table with Naïve Bayes’) can perform in abstracting
accurate yet comprehensible models from the Medical data.
Hence, We Propose –A Novice Predictive Modeling approach known as Rule Based Classification as an
alternative technique to decision trees, Association rules, Neural networks, Bayesian Networks in context of
Predictive analysis of Breast Cancer Recurrence .Further Owing the working principle of Rule Based
Classifiers and their advantages, we have generated a Rule Set which can predict Breast Cancer Recurrence
using the Patients’ Characteristics (attributes); thus Helping an Oncologist in Predicting Breast Cancer
Recurrence
2. Rule Based Classifiers
A Rule Based Classifier is a technique for classifying records using a collection of “If …. Then... ”
Rules .Figure1 shows an example of a Model generated by a Rule Based Classifier for a vertebrate
Classification problem .The Rules for the model are represented in Disjunctive Normal Form ,R=(r1 v r2
v ...rk),Where R is called as a Rule Set and ri’s are called Classification Rule or Disjuncts

13

Vol 2, No.2, 2012

Table 1. Example of Rule Set Vertebrate Classification Problem

r1:(gives birth=no )^ (Aerial creature=yes) Birds
r2:(gives birth=no) ^(Aquatic Creature =yes) Fishes
r3(Gives Birth=yes)^(BodyTemperature=Warm) Mammals
r4(Gives Birth =no) ^(Aerial Creature=no) Reptiles
r5:(Aquatic Creature=semi) Amphibians

2.1 Advantages of Rule Based Classifiers
Produce Descriptive Models ,which are easier to interpret, but gives comparable performance to
decision tree Classifiers

The Rule Based Classifiers create rectilinear partitions of attribute space and assign class to
each partition, nevertheless if Rule Based Classifier allows multiple rules to be triggered for a given
record ,then more complex decision boundary can be constructed

2.2 Working Principle
The Rule Based Classifier classifies a test record based on rule triggered by a record .To illustrate how a
Rule Based Classifier works ,Consider a rule set shown in the Table 2 and the following Vertebrates
Table 2 .Working Principle of Rule Based Classifiers

Body Temperature Skin Cover Gives Birth
Name Aquatic Creature

Lemur Warm Blooded Fur Yes No
Turtle Cold Blooded Scales No Semi
Dog Fish Cold Blooded Scales Yes Semi
Shark Cold Blooded Scales Yes Yes

The Working Principle of Rule Based Classifiers is shown in Table 2.A Rule Based Classifier classifies a test
record based on the rule triggered by the record .To illustrate how a Rule Based Classifier works ,Consider
the Rule Set shown in Table2 and the following vertebrates
• The First Vertebrate, which is a Lemur, is aWarm Blooded and gives birth to its young. It triggers
the Rule r3 ,and thus ,is Classified as a “Mammal”.
• The second vertebrate, which is a Turtle, triggers the rule r4 and r5, since the Classes predicted by
the rules are contradictory (reptiles versus amphibians),their conflicting classes must be resolved
• None of the Rules are applicable to Dogfish Shark. In this case ,we need to ensure that the
classifier can make a reliable prediction even though a teat record is not covered by any Rule
2.3 Advanced Rule Based Classifiers
Recently Number of Rule Based Classifiers has been proposed to get good Rule set; some of them are as
follows

14

Vol 2, No.2, 2012

2.3.1 Ripper
In our Experiment Java class implementing a Propositional Rule learner, RIPPER (Repeated Incremental
Pruning to produce Error reduction) which was proposed by William .W.Cohen is run against the Breast
Cancer Dataset taken from UCI Machine Learning Repository in WEKA Environment. The Algorithm is
briefly described as follows
Initialize RS = {}, and for each class from the less prevalent one to the more frequent one,
DO:
1. Building stage:
Repeat 1.1 and 1.2 until the description length (DL) of the rule set and examples is 64 bits greater than the
smallest DL met so far, or there are no positive examples, or the error rate >= 50%.
1.1. Grow phase:
Grow one rule by greedily adding antecedents (or conditions) to the rule until the rule is perfect (i.e. 100%
accurate). The procedure tries every possible value of each attribute and selects the condition with highest
information gain: p(log(p/t)-log(P/T)).
1.2. Prune phase:
Incrementally prune each rule and allow the pruning of any final sequences of the antecedents; The pruning
metric is (p-n)/(p+n) -- but it's actually 2p/(p+n) -1, so in this implementation we simply use p/(p+n)
(actually (p+1)/(p+n+2), thus if p+n is 0, it's 0.5).
2. Optimization stage:
After generating the initial rule set {Ri}, generate and prune two variants of each rule Ri from randomized
data using procedure 1.1 and 1.2. But one variant is generated from an empty rule while the other is
generated by greedily adding antecedents to the original rule. Moreover, the pruning metric used here is
(TP+TN)/(P+N).Then the smallest possible DL for each variant and the original rule is computed. The
variant with the minimal DL is selected as the final representative of Ri in the ruleset.After all the rules in
{Ri} have been examined and if there are still residual positives, more rules are generated based on the
residual positives using Building Stage again.
3. Delete the rules from the rule set that would increase the DL of the whole rule set if it were in it. and add
resultant rule set to RS
END DO
2.3.2 Decision Table
In our Experiment, A Java Class for building and using a simple decision table majority Classifier is run
against the Breast Cancer Dataset derived from UCI Machine Learning Repository. Given a training set of
labeled instances, an induction algorithm builds a classifier.There are two variants of decision table
classifiers based conceptually on a simple lookup table. The classifier, called DTMaj (DecisionTable
Majority) returns the majority of the training set if the decision table cell matching the new instance is
empty, i.e., it does not contain any training instances.
A decision table has two components- A schema, which is a list of attributes and a A body, which is a
multiset of labeled instances. Each instance consists of a value for each of the attributes in the schema and
a value for the label. The set of instances with the same values for the schema attributes is called a cell.
Given an unlabeled instance, ~x, the label assigned to the instance by a decision table classifier is computed
as follows. Let I be the set of labeled instances in the cell that exactly matches the given instance ~x, where
only the attributes in the schema are required to match and all other attributes are ignored. If I ≠0;; return
the majority class in I , breaking ties arbitrarily. Otherwise (I = 0;), A DTMaj returns the majority class in
the decision table ,can be implemented in a universal hash table

15

Vol 2, No.2, 2012

2.3.3 Decision Table with Naïve Bayes’(DTNB)
In our Experiment, A Java Class for building and using a decision table/naive bayes hybrid classifier is run
against Breast Cancer dataset derived from UCI Machine Learning Repository. The algorithm for learning
the combined model (DTNB) proceeds in much the same way as the one for stand-alone DTs. At each point
in the search it evaluates the merit associated with splitting the attributes into two disjoint subsets: one for
the DT, the other for NB. We use a forward selection, where, at each step, selected attributes are modeled
by NB and the remainder by the DT and all attributes are modeled by the DT initially. At each step, the
algorithm also considers dropping an attribute entirely from the model.
3. Proposed Methodology
The Breast Cancer Dataset obtained from UC- Irvine archive of Machine Repository has 201 instances of
One Class and 85 instances of other Class .The instances are described by 9 attributes some of which are
Linear and some are Nominal.
Attribute Information:
1.Class:no-recurrence-events,recurrence-events
2. age: 10-19, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79, 80-89,90-99.
3.menopause:lt40,ge40,premeno.
4. tumor-size: 0-4, 5-9, 10-14, 15-19, 20-24, 25-29, 30-34, 35-39,40-44,45-49,50-54,55-59.
5.inv-nodes:0-2,3-5,6-8,9-11,12-14,15-17,18-20,21-23,24-26,27-29,30-32,33-35,36-39.
6.node-caps:yes,no.
7.deg-malig:1,2,3.
8.breast:left,right.
9. breast-quad: left-up, left-low, right-up, right-low, central.
10. Irradiat: yes, no.
The Proposed Methodology for the automation of Breast Cancer Health Records is depicted in fig 1. The
Experiments are carried out as depicted in the fig1. Initially Breast Cancer Data Set Collected from UCI
Machine Learning Repository is Pre Processed to Remove Missing Values in order to make it suitable for
Mining Process. Once the Pre Processing gets over ,Rule Based Classifiers like DT,DTNB,RIPPER are
applied one by one, and accuracy is measured .DTNB gives better accuracy compared to other two rule
based classifiers .Hence the Rule Set generated by DTNB is stored in MS-Access Database
A New patient is labeled by validating the Rule set shown in Table 7 generated by DTNB and by using the
working principle of Rule Based Classifier and Patients’ Characteristics.

Figure 1. .Proposed Methodology for Breast Cancer Recurrence Detection

16

Vol 2, No.2, 2012

4. Experimental Results
The Classification Accuracy of the three classifiers run against Breast Cancer dataset is shown in Table 3
Table 3 -Classification Accuracy of RIPPER, DT, and DTNB
Sl no Dataset RIIPER DT DTNB

1 Breast Cancer 72.27% 72.72% 75.17%

The Confusion Matrices of RIPPER, DT, and DTNB, when run against Breast Cancer Dataset are shown in
Table 4, 5, 6 respectively. Table 7 gives the overview of Rule Set Generated by DTNB.
We Employed four Performance Measures –Precision, Recall-Measure and ROC Space .A distinguished
Confusion Matrix (Contingency Table) is obtained to calculate the four measures .It contains ,The cell
which denotes the number of Samples Classified as true while they were true(TP) and the Cell which
denotes the number of Samples Classified as False while they were False(TN) . The other two cells denote
the Number of Samples Misclassified –Specifically the Cell denoting Number of Samples Classified as
False while they were actually true (FN) and the Cell denoting Number of Samples Classified as True
while they were actually False (FP) .Once the Confusion Matrix is constructed ,The Precision Recall and
F-Measure are calculated as follows.
Recall=TP/ (TP+FN) (1)
Precision=TP/ (TP+FP) (2)
F-Measure= (2*TP)/ (2*TP+FP+FN) (3)
Precision measures percentage of actual Patients (TP) among patients that get declared disease .Recall
measures percentage of actual patients that were discovered .F-Measure balances between Precision and
Recall. A ROC (Receiver Operator Characteristic) space is defined by FP Rate and TPR Rate as X and
Y Axes respectively
TPR=TP/ (TP+FN) (4)

FPR=FP/ (FP+TN) (5)
Table 4: Confusion Matrix of RIPPER
TP Rate FP Rate Precision Recall F-Measure ROC - Class
Area

0.846 0.635 0.759 0.846 0.8 0.589 No-Recurrence-Events
0.365 0.154 0.5 0.365 0.422 0.589 Recurrence-Events

Table 5: Confusion Matrix of DT
TP FP Precision Recall F-Measure ROC-Area Class
Rate Rate


17

Vol 2, No.2, 2012

Table 6: Confusion Matrix of DTNB
TP Rate FP Rate Precision Recall F-Measure ROC Area Class


Table 7: Rule Set of DTNB
age menopause tumor-size inv-nodes breast breast-quad Class
50-59 premeno 35-39 6-8 right right up no-recurrence-events
60-69 ge40 15-17 15-19 left right-low no-recurrence events
40-49 premeno 45-49 12-14 right right up no-recurrence events

5. Conclusion and Future Work
In this paper, we have discussed need for Data mining in the Medical field. In this context we have
discussed a New Predictive Modeling approach known as Rule Based Classification as an alternative
technique to Neural Networks, Bayesian Networks, Decision Trees and Association Rules along with its
advantages and working principle. We have assessed the performance of three rule based classifiers namely
DT, RIIPER and DTNB and found that classification accuracy of DTNB is better compared to other two
classifiers, when run against Breast Cancer Dataset obtained from UCI Machine learning Repository.
Further we have found Rule set containing some interesting rules which were easy to interpret and familiar
to represent them in spreadsheet were obtained from DTNB.The Experimental results also reveal that
DTNB is an efficient approach for extraction of patterns from Breast cancer dataset. In the Future Work,
Decision Table can be combined with Genetic Algorithm to reduce the attributes for the prediction of Breast
Cancer Recurrence through a Graphical User Interface or -Fuzzy Weighted Association Rule Mining
Technique may be used along with a GUI and may be used to label new patients as developing or not
Breast Cancer Recurrence Events
References
Cleveland Clinical Database http:// www.clevelandclinic.org/health
Feelders, A., Daniels, H. and Holsheimer, M. (2000) ‘Methodological and Practical Aspects of Data
Mining’, Information and Management, 271-281
UCI Machine Laening Repository http://www.ics.uci.edu/`mlax/MLrepository.html
Belciug,S. Gorunescu,F. Salem ,A,-B.Gorunsecu ,M.(2010).,”Clustering Based Approach for detecting
Breast Cancer Recurrence” .In International Conference on Intelligent system Design and Application
(ISDA) ,533-538
Jiawei Han, Micheline Kamber, and Jian Pei (2001), “Data Mining: Concepts and Techniques “3rd edition,
Morgan Kaufmann publication (Chapter 1)
William W. Cohen(1995): “Fast Effective Rule Induction.” In: Twelfth International Conference on
Machine Learning, 115-123
Ron Kohavi(1995): The Power of Decision Tables. In: 8th European Conference on Machine Learning,
174-189

18

Vol 2, No.2, 2012

Mark Hall, Eibe Frank: Combining Naive Bayes and Decision Tables(2008), In: Proceedings of the 21st
Florida Artificial Intelligence Society Conference (FLAIRS),318-319
D.SenthilKumar ,G.SathyaDevi and S.Sivanesh(May 2011) “Decision Support System for Medical
Diagnosis using Data Mining ”In International Journal of Computer Science Issues(IJCSI) , Vol 3, Issue 3,
No.1,147-153
H.W. Ian ,E.F(2005), “Data Mining Practical Machine Learning Tools and Techniques ”,Morgan Kaufman
publication

First Author Srinivas Murti received the B.E degree in Information Science and Engineering from
Vishweswarayya Technological University (VTU), Belgaum (INDIA) in 2008.He is currently an M.Tech
(QIP) Candidate in Department of Computer Science and Engineering, PG Extension Centre, VTU,
Bagalkot.He has completed 2.5 Years of teaching in various courses of Undergraduate Engineering
Programs. He has Industrial Experience of 1 year in Software Development and Maintainance.His Research
Interests include Medical Data Mining, Machine learning and related Real world applications and published
various research papers in leading journals and conferences
Second Author Mahantappa is currently an M.Tech (QIP) Candidate in Department of Computer Science
and Engineering, PG Extension Centre, VTU, Bagalkot.He has completed 2.5 Years of teaching in various
courses of Undergraduate Engineering Programs. His Research interests include Image processing, Radio
Frequency Identification Devices (RFID) and Technology, Networking and Data Mining

19

International Journals Call for Paper
The IISTE, a U.S. publisher, is currently hosting the academic journals listed below. The peer review process of the following journals
usually takes LESS THAN 14 business days and IISTE usually publishes a qualified article within 30 days. Authors should
send their full paper to the following email address. More information can be found in the IISTE website : www.iiste.org

Business, Economics, Finance and Management PAPER SUBMISSION EMAIL
European Journal of Business and Management EJBM@iiste.org
Research Journal of Finance and Accounting RJFA@iiste.org
Journal of Economics and Sustainable Development JESD@iiste.org
Information and Knowledge Management IKM@iiste.org
Developing Country Studies DCS@iiste.org
Industrial Engineering Letters IEL@iiste.org

Physical Sciences, Mathematics and Chemistry PAPER SUBMISSION EMAIL
Journal of Natural Sciences Research JNSR@iiste.org
Chemistry and Materials Research CMR@iiste.org
Mathematical Theory and Modeling MTM@iiste.org
Advances in Physics Theories and Applications APTA@iiste.org
Chemical and Process Engineering Research CPER@iiste.org

Engineering, Technology and Systems PAPER SUBMISSION EMAIL
Computer Engineering and Intelligent Systems CEIS@iiste.org
Innovative Systems Design and Engineering ISDE@iiste.org
Journal of Energy Technologies and Policy JETP@iiste.org
Information and Knowledge Management IKM@iiste.org
Control Theory and Informatics CTI@iiste.org
Journal of Information Engineering and Applications JIEA@iiste.org
Industrial Engineering Letters IEL@iiste.org
Network and Complex Systems NCS@iiste.org

Environment, Civil, Materials Sciences PAPER SUBMISSION EMAIL
Journal of Environment and Earth Science JEES@iiste.org
Civil and Environmental Research CER@iiste.org
Civil and Environmental Research CER@iiste.org

Life Science, Food and Medical Sciences PAPER SUBMISSION EMAIL
Journal of Biology, Agriculture and Healthcare JBAH@iiste.org
Food Science and Quality Management FSQM@iiste.org
Chemistry and Materials Research CMR@iiste.org

Education, and other Social Sciences PAPER SUBMISSION EMAIL
Journal of Education and Practice JEP@iiste.org
Journal of Law, Policy and Globalization JLPG@iiste.org Global knowledge sharing:
New Media and Mass Communication NMMC@iiste.org EBSCO, Index Copernicus, Ulrich's
Journal of Energy Technologies and Policy JETP@iiste.org Periodicals Directory, JournalTOCS, PKP
Historical Research Letter HRL@iiste.org Open Archives Harvester, Bielefeld
Academic Search Engine, Elektronische
Public Policy and Administration Research PPAR@iiste.org Zeitschriftenbibliothek EZB, Open J-Gate,
International Affairs and Global Strategy IAGS@iiste.org OCLC WorldCat, Universe Digtial Library ,
Research on Humanities and Social Sciences RHSS@iiste.org NewJour, Google Scholar.

Developing Country Studies DCS@iiste.org IISTE is member of CrossRef. All journals
Arts and Design Studies ADS@iiste.org have high IC Impact Factor Values (ICV).

11.using rule based classifiers for the predictive analysis of breast cancer recurrence

More Related Content

What's hot (20)

Similar to 11.using rule based classifiers for the predictive analysis of breast cancer recurrence (20)

More from Alexander Decker (20)

Recently uploaded (20)

11.using rule based classifiers for the predictive analysis of breast cancer recurrence