Breast cancer prediction system using Data mining methods · 2018-05-06 · Breast cancer prediction system using Data mining methods Dr. C Nalini 1 D.Meera 2 1Professor, 2Mtech Student,

Breast cancer prediction system using Data mining methods

Dr. C Nalini1 D.Meera

2

1Professor,

2Mtech Student, Computer Science and Engineering,

BIST, BIHER,BharathUniversity,Chennai – 73.

[email protected]

ABSTRACT

Cancer is the second most common cause of death worldwide with an estimated 7.9 million of

deaths in 2007. This number is projected to rise further and reach 12 million deaths in 2030

which makes cancer a major public health issue. Data mining is defined as a process used to

extract usable data from a large data set. There are many different areas under data mining and

one of them is classification or the supervised learning. In the health care industry, the data

mining is predominantly used for disease prediction.Classification also can be implemented

through a number of different approaches or algorithms.There are many different areas under

Data Mining and one of them is Classification or the supervised learning.We have conducted the

comparison between two algorithms with help of WEKA (The Waikato Environment for

Knowledge Analysis), which is an open source software. It contains different type’s data mining

algorithms.This paper explains discussion of Bayesian Network and J48 algorithms. Here, for

comparing the result, we have used the analysis of classification accuracy and execution time

and various parameters.

KEYWORDS

Classification, Decision tree [J48], Naïve Bayes, Breast cancer

1. INTRODUCTION

The second leading cause of death among women is breast cancer and it comes directly after

lung cancer [1-9]. The health and medical sector is more in need of data mining today. When

certain data mining methods used, valuable information can be extracted from large data base

and that can help to medical practitioner to take decision, and improve health services. There are

a few arguments that can support the use of data mining in health sector for breast cancer like

early detection, early avoidance, and indication based medication, rectifying hospital data errors

[10-15].WEKA is a powerful tool as it contains supervised learning as well unsupervised

learning methods. It contains Classification, Clustering, Association Mining, Feature Selection,

Data Visualization, etc. The main reason behind using WEKA is helps researchers like us to

implement and compare data mining techniques very easily on real or synthetic data. It is also

well suited for developing new machine learning techniques[5]. WEKA comes under the open

source software issued under GNU General Public License[16-21].

International Journal of Pure and Applied MathematicsVolume 119 No. 12 2018, 10901-10911ISSN: 1314-3395 (on-line version)url: http://www.ijpam.euSpecial Issue ijpam.eu

10901

Breast cancer is an uncontrolled growth of breast cells. It refers to a malignant tumor that has

developed from cells in the breast. Breast Cancer constitutes a major public health issue globally

with over 1 million new cases diagnosed annually,resulting in over 400,000 annual deaths and

about4.4 million women living with the disease. It alsoaffects one in eight women during their

lives. It isthe commonest site specific malignancy affectingwomen and the most common cause

of cancer mortality in women worldwide. It is also found in men but not very common.

Statistics available in Nigeria are largely unreliable because of many factors that have not

allowed adequate data collection and documentation; but according to DrChinyereAkpanika,

[22-29] a Gynaecologist at the University of Calabar Teaching Hospital (UCTH) said Nigeria

recorded over 100,000 new cases of cancer annually, added that early detection of the disease

increased the chance of survival of an affected person. According to her, 85 per cent of women

who have breast cancer do not have a family history of breast cancer. The Medical

Director,Optimal Cancer Care Foundation, Dr Femi Olaleye, [2] said breast cancer killed one in

every 25 Nigerian women.

Breast cancer is a malignant or benign tumor, inside breast, wherein cellsdivide and grow

without control [30-34]. Scientists have tried to know the exact reason behind breast cancer, as

there are a few risk factors which increase the like hood of a woman developing breast cancer.

Age, genetic risk and family history are some such factors being considered for breast

cancer[3].Treatments of breast cancer are divided into two types, local and systematic. Surgery

and radiation are local type of treatments whereas chemotherapy and hormone therapy are

examples of systematic therapies. For getting best results, both treatments are used together in

different variations as per the patient and disease intensity[35-39].These records are cleaned and

filtered with the intention that the irrelevant data from the warehouse would be removed before

mining process occurs.

The aim of this paper is to predict the Breast Cancer effectively by applying the attribute subset

evaluator namely Consistency Subset Evaluator.Usingthese significant attributes the

classification of dataset is performed using naïve bayesand J48. The remaining portion of the

paper is discussed as follows. The proposed methodology is given in Section 3. Section [4]

analyses the experimental results. Section [40-45] gives conclusion.

2. LITERATURE SURVEY: Dr.Rajesh et al (2012) who used SEER dataset for the diagnosis of breast cancer using the C4.5

classification algorithm. The algorithm was used to classify patients into either pre-cancer stage

or potential breast cancer cases. During the testing phase, the C4.5 classification rules were

applied to a test sample and the algorithm showed had an accuracy of 92.2%, sensitivity of

46.66% and a specificity of 97.4%. Future enhancement of the work will require the

improvisation of the C4.5 algorithm to improve classification rate to achieve greater accuracy.

Shajahan et al (2013) worked on the application of data mining techniques to model breast

cancer data using decision trees to predict the presence of cancer. Data collected contained 699

instances (patient records) with 10 attributes and the output class as either benign or malignant.

Input used contained sample code number, clump thickness, cell size and shape uniformity, cell

growth and other results physical examination[6]. The results of the supervised learning

International Journal of Pure and Applied Mathematics Special Issue

10902

algorithm applied showed that the random tree algorithm had the highest accuracy of 100% and

error rate of 0 while CART had the lowest accuracy with a value of 92.99% but naïve bayes’ had

the an accuracy of 97.42% with an error rate of 0.0258.

Chaurasia and Pal6 offered three popular data mining algorithms: CART, ID3 (iterative

dichotomized 3), and DT for diagnosing heart diseases, and the results presented demonstrated

that CART obtained higher accuracy within less time.

Chaurasia and Pal5 compare the performance criterion of supervised learning classifiers, such

as Naıve Bayes, SVM-RBF kernel, RBF neural networks, Decision Tree (Dt) (J48), and simple

classification and regression tree (CART), to find the best classifier in breast cancer datasets. The

experimental result shows that SVM-RBF kernel is more accurate than other classifiers; it scores

at the accuracy level of 96.84% in the Wisconsin Breast Cancer (original) datasets.

Qi Fanet.al [6] the authors of this paper focus on SEER public use-data to predict Brest Cancer.

They use pre-classification method and find a possible solution to discover thein formation of

Brest Cancer.H.S.Hota [2] built a classification model using various intelligent techniques such

as ANN(Artificial Neural Network),Unsupervised Artificial Neural Network, Statistical

technique and decision tree. Experimental results show a testing accuracy of 97.73% from which

the efficiency of the ensemble model was highlighted.

Ryan Potter has performed the comparison of classification algorithms on breast cancer dataset

to perform the diagnosis of the patients.

3. METHODOLOGY: 3.1 Dataset:

The breast cancer dataset have been created for analysis of diabetic disease.This dataset contains

seven hundred and sixty eight instances and eight attributes are used in this comparative analysis.

Menstrual and reproductive history:

Earlymenstruation (before age 12), late menopause(after age 55), having your first child at an

olderage, or never having given birth can alsoincrease the risk of breast cancer.

Certain genome:

This is cause by mutation incertain genes and can be determined throughgenetic test as

individual withcertain gene mutation can pass it onto their children.

Dense breast tissue:

This can increase the risk of having breast cancer and makes lumps harder to detect.

Population

This study was performed on TMA images from 164 fe- male patients with an invasive ductal

carcinoma of the breast, treated at the Oncology Institute of Vilnius University and investigated

at the National Center of Pathology,during the period of 2007 to 2009. The study was

approvedby the Lithuanian Bioethics Committee.The patients’ con-sent to participate in the

study was obtained.

Tissue preparation

The TMAs were constructed, stained and scanned as described previously [3]. Briefly,

onemillimetre-diameter cores were punched from tumour areas randomly selected by the

pathologist and paraffin sections were cut at 3μm-thickness[3].


10903

Immunohistochemistry (IHC)

IHC for Ki67 was performed with a multimer-technology based detection system, ultraView

Universal DAB (Ventana, Tucson, AZ, USA). The Ki67 antibody (clone MIB-1;DAKO,

Glostrup, DK) was applied at a 1:200 dilution for 32 minutes, followed by the

VentanaBenchMark XT automated immunostainer (Ventana) standard Cell Conditioner1 (CC1, a

proprietary buffer) at 95°C for 64 minutes[5].Finally, the sections were developed in DAB at

37°C for eight minutes, counterstained with Mayershematoxylin and mounted.

Image acquisition

Digital images were captured using the Aperio Scan Scope XT Slide Scanner(Aperio

Technologies, Vista,CA, USA) under 20x objective magnification (0.5μm resolution). One TMA

spot image per patient was used for the study.

Quantification with stereology test grid

RV were obtained by marking Ki67-positive and negativetumour cell profiles, using a

stereological method for 2Dobject enumeration [4,5] implemented by the Stereology module

(ADCIS, Caen, France) with a test grid ofsystematically sampled frames (frame size - 125

pixels,spacing of frames - 250 pixels) overlaid on a spot imagein ImageScope (Aperio

Technologies, USA),

Visual evaluation (VE)

A global subjective impression for the Ki67 LI on the sameimages was performed by five

pathologists independentlyand provided semi-quantitative values (Ki67-VE-1, 2, 3, 4and 5)

expressed as the percentage of Ki67-positivetumour cell profiles. Counting was not included in

theprocedure.

3.2 Classification

Classification is a process that is used for partitioning the data into different classes according to

some constrains or it classify each item in a dataset into one of predefined set of classes or

groups. It is a supervised learning approach having known class categories.Several major kinds

of classification algorithms including C4.5, J48, k-nearest neighbor classifier, Naive Bayes,

SVM, Apriori, and AdaBoost. In this research work Naïve Bayes, decision tree[J48] algorithms

are used to predict the breast cancer disease.

3.2.1 Naive Bayes Naive Bayes is a machine learning algorithm for classification problems. It is based on Thomas

Bayes’s probability theorem. It is primarily used for text classification which involves high-

dimensional training datasets. A few examples are spam filtration, sentimental analysis, and

classifying news articles. It is not only known for its simplicity but also for its effectiveness.

It is fast to build models and make predictions with Naive Bayes algorithm. Naive Bayes

algorithm is the algorithm that learns the probability of an object with certain features belonging

to a particular group/class.In short, it is a probabilistic classifier. The Naive Bayesalgorithm is

called “naive” because it makes theassumption that the occurrence of a certain feature

isindependent of the occurrence of other features. The“Bayes” part refers to the statistician and

philosopherThomas Bayes and the theorem was named after him,Baye’s theorem, which is the

base for Naı¨ve Bayes algorithm.Naive Bayes algorithm is Baye’s theorem oralternatively known

as Baye’s rule or Baye’s law. Itgives us a method to calculate the conditional probability,i.e. the

probability of an event based on previousknowledge available on the events. More

formally,Baye’s theorem is stated as the following equation


10904

P(A/B)=P(B/A)P(A) / P(B)

• P(A/B): Probability (conditional probability) ofoccurrence of event A given the event B is true.

• P(A) and P(B): Probabilities of the occurrence ofevent A and B, respectively.

• P(B/A): Probability of the occurrence of event Bgiven the event A is true.

3.2.2 Decision Trees (J48)

J48 decision trees classifier is a simple decision learning algorithm, it accepts only categorical

data for building a model. The basic idea of ID3 is to construct a decision tree by employing a

top down greedy search through the given sets of training data to test each attribute at every

node. It uses statistical property known as information gain to select which attribute to test at

each node in the tree. Information gain measures how well a given attribute separates the training

samples according to their classification.

It is suitable for handling both categorical as well as continuous data. A threshold value is fixed

such that all the values above the threshold are not taken into consideration. The initial step is to

calculate information gain for each attribute. The attribute with the maximum gain will be

preferred as the root node for the decision tree.

Given a set S of breast cancer cases, J48 first grows an initial tree using the divide-and-conquer

algorithm as follows:

• If all the cases in S belong to the same class or S is small, the tree is a leaf labeled with the

most frequent class in S;

• Otherwise, choose a test based on a single attribute with two or more outcomes. Make this test

the root of the tree with one branch for each outcome of the test, partition S into corresponding

subsets S1, S2,……, Sn for a dataset containing n cases according to the outcome for each case,

and apply the same procedure recursively to each subset. It uses a statistical property known as

information gain to select which attribute to test at each node in the tree. It measures how well a

given attribute separates the training samples according to their classification.

4EXPERIMENTAL RESULTS

This work is implemented in Weka tool. Weka is a collection of machine learning algorithms for

data mining tasks. The algorithms can either be applied directly to a dataset or called from your

own Java code. Weka contains tools for data pre-processing, classification, regression,clustering,

association rules, and visualization. It is also well-suited for developingnew machine learning

schemes.The experimental comparison of classification algorithms are done based on the

performance measures of classification accuracy and execution time.

4.1 Classification Accuracy Accuracy:

Accuracy is defined in the terms of correctly classified instances divided by the total

number of instances present in the dataset.

Accuracy = 𝑇𝑃+𝑇𝑁

𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁

Where TP- True Positive, FP- False Positive, TN- True Negative, FN- False Negative

TP Rate:


10905

It is the ability which is used to find the high true-positive rate. The true-positive rate is

also called as sensitivity.

TPR= 𝑇𝑃

𝑇𝑃+𝐹𝑁

Precision:

Precision is given the correlation of number of modules correctly classified to the number

of entire modules classified fault-prone. It is quantity of units correctly predicted as faulty.

Precision= 𝑇𝑃

𝑇𝑃+𝐹𝑃

F-Measure:

F- Measure is the one has the combination of both precision and recall which is used to

compute the score.

F-Measure= 2∗𝑟𝑒𝑐𝑎𝑙𝑙 ∗𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛

𝑟𝑒𝑐𝑎𝑙𝑙 +𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛

Table: Accuracy Measure for Classifier Algorithms

Algorithms Correctly

classified

instances

(%)

Incorrectly

classified

instances

(%)

TP rate F

Measure

IR

precision

IR

Recall

Accuracy

(%)

Naïve bayes 215 71 0.85 0.81 0.78 0.85 64%

J48 217 69 0.95 0.84 0.75 0.95 60%

From the table, In performance measures Decision tree(j48)performs best in classifying process

than Naïve Bayes algorithm and In accuracy naïve bayes algorithm is better than J 48.

In explorer:

Figure : decision tree


10906

In knowledge flow environment:

4.2: Execution time

Algorithms Execution time (sec)

Naïve bayes 0

J48 0.1

From the table naïve Bayes takes less execution time than the J48 algorithm.

5. RESULT AND DISCUSSION

The simulation results are partitioned into several sub items for easier analysis and evaluation.

The algorithm will be compared by analyzing their performance; execution time.The algorithm

which has the higher accuracy with the minimum execution time has chosen as the best

algorithm. In this classification, naïve bayes has the maximum classification accuracy,

performance factors and minimum execution time. So it is considered as the best classification

algorithm. . From the results , it is very clear that data mining techniques can be used in

predicting breast cancer risks and that the J48 decision trees has a better accuracy than the naïve

bayes’ model which is a statistical tool.

6. CONCLUSION

In this paper, we applied two prediction models for breast cancer. Here, we used two popular

data mining methods: Naive Bayes and J48. In this research work classification process is used

to classify four types of kidney diseases. Comparison of J48 and Naïve Bayes classification

algorithms is done based on the performance factors classification accuracy and execution time.

From the results, the Naïve bayesthat are accurate, hence it is considered as best classifier when

compared with J48 algorithm. Perhaps, Naïve Bayes classifier classifies the data with minimum

execution time.


10907

REFERENCES

1. Hameed Hussain, J., Sharavanan, R., Floor cleaning machine by remote control,

AInternational Journal of Pure and Applied Mathematics, V-116, I-14 Special Issue, PP-

461-464, 2017

2. Hameed Hussain, J., Srinivasan, V., Extraction of polythene waste from domestic waste,

International Journal of Pure and Applied Mathematics, V-116, I-14 Special Issue, PP-

427-431, 2017

3. Hameed Hussain, J., Thirumavalavan, S., Flow analysis of copper tube for solar trough

collector without joint, International Journal of Pure and Applied Mathematics, V-116, I-

14 Special Issue, PP-541-544, 2017

4. Hanirex, D.K., Kaliyamurthie, K.P., Mining the financial multi-relationship with accurate

models, Middle - East Journal of Scientific Research, V-19, I-6, PP-795-798, 2014

5. Hemapriya, M., Meikandaan, T.P., Repair of damaged reinforced concrete beam by

externally bonded with CFRP sheets, International Journal of Pure and Applied

Mathematics, V-116, I-13 Special Issue, PP-473-479, 2017

6. Hemapriya, M., Meikandaan, T.P., Experimental study on changes in properties of

cement concrete using steel slag and fly ash, International Journal of Pure and Applied


7. Hemapriya, M., Meikandaan, T.P., Experimental study on structural repair and

strengthening of RC beams with FRP laminates, International Journal of Pure and

Applied Mathematics, V-116, I-13 Special Issue, PP-355-361, 2017

8. Hemapriya, M., Meikandaan, T.P., Effect of high range water reducers on sorptivity and

water permeability of concrete, International Journal of Pure and Applied Mathematics,

V-116, I-13 Special Issue, PP-377-381, 2017

9. Hemapriya, M., Meikandaan, T.P., Strength and workability characteristics of super

plasticized concrete, International Journal of Pure and Applied Mathematics, V-116, I-13

Special Issue, PP-345-353, 2017

10. Hemapriya, M., Meikandaan, T.P., Potency and workability behavior of quality

plasticized structural material, International Journal of Pure and Applied Mathematics, V-

116, I-13 Special Issue, PP-363-367, 2017

11. Hussain, J.H., Manavalan, S., Optimization of properties of jatropha methyl Ester (JME)

from jatropha oil, International Journal of Pure and Applied Mathematics, V-116, I-18


12. Hussain, J.H., Manavalan, S., Optimization and comparison of properties of neem and

jatropha biodiesels, International Journal of Pure and Applied Mathematics, V-116, I-17


13. Hussain, J.H., Meenakshi, C.M., Simulation and analysis of heavy vehicles composite

leaf spring, International Journal of Pure and Applied Mathematics, V-116, I-17 Special

Issue, PP-135-140, 2017

14. Hussain, J.H., Nimal, R.J.G.R., Review: Investigation on mechanical properties of

different metal matrix composites in diffusion bonding method by using metal


10908

interlayers, International Journal of Pure and Applied Mathematics, V-116, I-18 Special

Issue, PP-459-464, 2017

15. Jagadeeswari, P., Subashini, G., Basic results of probability, International Journal of Pure

and Applied Mathematics, V-116, I-17 Special Issue, PP-275-276, 2017

16. Janani, V.D., Kavitha, S., Conceptual level similarity measure based review spam

detection adversarial spam detection using the randomized hough transform-support

vector machine, International Journal of Pure and Applied Mathematics, V-116, I-9


17. Jasmin, M., Beulah Hemalatha, S., Security for industrial communication system using

encryption / decryption modules, International Journal of Pure and Applied Mathematics,

V-116, I-15 Special Issue, PP-563-567, 2017

18. Jasmin, M., Beulah Hemalatha, S., VLSI-based frequency spectrum analyzer for low area

chip design by using yasmirub method, International Journal of Pure and Applied


19. Jasmin, M., Beulah Hemalatha, S., RFID security and privacy enhancement, International

Journal of Pure and Applied Mathematics, V-116, I-15 Special Issue, PP-535-538, 2017

20. Jasmin, M., Beulah Hemalatha, S., Digital phase locked loop, International Journal of

Pure and Applied Mathematics, V-116, I-15 Special Issue, PP-569-574, 2017

21. Jeyalakshmi, G., Arulselvi, S., Community oriented configurations for WSN,


529-533, 2017

22. Jeyalakshmi, G., Arulselvi, S., Investigating file systems, International Journal of Pure

and Applied Mathematics, V-116, I-15 Special Issue, PP-517-521, 2017

23. Jeyalakshmi, G., Arulselvi, S., Methodology for the development of lambda calculus,


511-515, 2017

24. Jeyalakshmi, G., Arulselvi, S., Remote procedure calls in access points, International


25. Jeyanthi Rebecca, L., Anbuselvi, S., Sharmila, S., Medok, P., Sarkar, D., Effect of marine

waste on plant growth, Der Pharmacia Lettre, V-7, I-10, PP-299-301, 2015

26. Kaliyamurthie, K.P., Parameswari, D., Udayakumar, R., Malicious packet loss during

routing misbehavior-identification, Middle - East Journal of Scientific Research, V-20, I-

11, PP-1413-1416, 2014

27. Kanagavalli, G., Sangeetha, M., Intelligent trafficlight system for reducedfuel

consumption, International Journal of Pure and Applied Mathematics, V-116, I-15


28. Kanagavalli, G., Sangeetha, M., GPS based blind pedestrian positioning and voice

response system, International Journal of Pure and Applied Mathematics, V-116, I-15


29. Kanagavalli, G., Sangeetha, M., Detection of retinal abnormality by contrast

enhancement methodusingcurvelet transform, International Journal of Pure and Applied



10909

30. Kanagavalli, G., Sangeetha, M., Design of low power VLSI circuits for precharge logic,


505-509, 2017

31. Kanniga, E., Selvaramarathnam, K., Sundararajan, M., Kandigital bike operating system,

Middle - East Journal of Scientific Research, V-20, I-6, PP-685-688, 2014

32. Karthik, B., Arulselvi, Noise removal using mixtures of projected gaussian scale

mixtures, Middle - East Journal of Scientific Research, V-20, I-12, PP-2335-2340, 2014

33. Karthik, B., Arulselvi, Selvaraj, A., Test data compression architecture for lowpowervlsi

testing, Middle - East Journal of Scientific Research, V-20, I-12, PP-2331-2334, 2014

34. Karthikeyan, R., Michael, G., Kumaravel, A., A housing selection method for design,

implementation&evaluation for web based recommended systems, International


35. Khanaa, V., Thooyamani, K.P., Using lookup table circulating fluidised bed combustion

boiler by the method of sensor output linearization, Middle - East Journal of Scientific

Research, V-16, I-12, PP-1801-1806, 2013

36. Khanaa, V., Thooyamani, K.P., Face routing protocol using genetic algorithm in, Middle

- East Journal of Scientific Research, V-16, I-12, PP-1863-1867, 2013

37. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Two factor authentication using mobile

phones, World Applied Sciences Journal, V-29, I-14, PP-208-213, 2014

38. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Next major wave of it inovation, World

Applied Sciences Journal, V-29, I-14, PP-218-220, 2014

39. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Traffic policing approach for wireless

video conference traffic, World Applied Sciences Journal, V-29, I-14, PP-200-207, 2014

40. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Patient monitoring in gene ontology

with words computing using SOM, World Applied Sciences Journal, V-29, I-14, PP-195-

199, 2014

41. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Load balancing in structured PEER to

PEER systems, World Applied Sciences Journal, V-29, I-14, PP-186-189, 2014

42. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Impact of route stability under random

based mobility model in MANET, World Applied Sciences Journal, V-29, I-14, PP-274-

278, 2014

43. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Modelling Cloud Storage, World

Applied Sciences Journal, V-29, I-14, PP-190-194, 2014

44. Khanaa, V., Thooyamani, K.P., Udayakumar, R., Elliptic curve cryptography using in

multicast network, World Applied Sciences Journal, V-29, I-14, PP-264-269, 2014

45. Khanaa, V., Thooyamani, K.P., Udayakumar, R., SRW/U as a lingua franca in managing

the diversified information resources, World Applied Sciences Journal, V-29, I-14, PP-

279-284, 2014


10910

10911

10912

Documents

Breast cancer prediction system using Data mining methods · 2018-05-06 · Breast cancer prediction system using Data mining methods Dr. C Nalini 1 D.Meera 2 1Professor, 2Mtech Student,