Analysis of cancer data: A data mining approach

Research output: Contribution to journalArticle

51 Citations (Scopus)

Abstract

Even though cancer research has traditionally been clinical and biological in nature, in recent years data driven analytic studies have become a common complement. In medical domains where data and analytics driven research is successfully applied, new and novel research directions are identified to further advance the clinical and biological studies. In this research, we used three popular data mining techniques (decision trees, artificial neural networks and support vector machines) along with the most commonly used statistical analysis technique logistic regression to develop prediction models for prostate cancer survivability. The data set contained around 120 000 records and 77 variables. A k-fold cross-validation methodology was used in model building, evaluation and comparison. The results showed that support vector machines are the most accurate predictor (with a test set accuracy of 92.85%) for this domain, followed by artificial neural networks and decision trees.

Original languageEnglish
Pages (from-to)100-112
Number of pages13
JournalExpert Systems
Volume26
Issue number1
DOIs
StatePublished - 1 Feb 2009
Externally publishedYes

Fingerprint

Decision tree
Data mining
Artificial Neural Network
Support Vector Machine
Cancer
Data Mining
Survivability
Prostate Cancer
Test Set
Logistic Regression
Decision trees
Data-driven
Cross-validation
Prediction Model
Statistical Analysis
Support vector machines
Predictors
Fold
Complement
Neural networks

Keywords

  • Cancer
  • Classification
  • Data mining
  • Decision trees
  • Neural networks
  • SEER
  • Sensitivity analysis

Cite this

@article{e47b1ca82ca44fbda57d9eed7d59ccf8,
title = "Analysis of cancer data: A data mining approach",
abstract = "Even though cancer research has traditionally been clinical and biological in nature, in recent years data driven analytic studies have become a common complement. In medical domains where data and analytics driven research is successfully applied, new and novel research directions are identified to further advance the clinical and biological studies. In this research, we used three popular data mining techniques (decision trees, artificial neural networks and support vector machines) along with the most commonly used statistical analysis technique logistic regression to develop prediction models for prostate cancer survivability. The data set contained around 120 000 records and 77 variables. A k-fold cross-validation methodology was used in model building, evaluation and comparison. The results showed that support vector machines are the most accurate predictor (with a test set accuracy of 92.85{\%}) for this domain, followed by artificial neural networks and decision trees.",
keywords = "Cancer, Classification, Data mining, Decision trees, Neural networks, SEER, Sensitivity analysis",
author = "Dursun Delen",
year = "2009",
month = "2",
day = "1",
doi = "10.1111/j.1468-0394.2008.00480.x",
language = "English",
volume = "26",
pages = "100--112",
journal = "Expert Systems",
issn = "0266-4720",
publisher = "Wiley-Blackwell Publishing Ltd",
number = "1",

}

Analysis of cancer data : A data mining approach. / Delen, Dursun.

In: Expert Systems, Vol. 26, No. 1, 01.02.2009, p. 100-112.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Analysis of cancer data

T2 - A data mining approach

AU - Delen, Dursun

PY - 2009/2/1

Y1 - 2009/2/1

N2 - Even though cancer research has traditionally been clinical and biological in nature, in recent years data driven analytic studies have become a common complement. In medical domains where data and analytics driven research is successfully applied, new and novel research directions are identified to further advance the clinical and biological studies. In this research, we used three popular data mining techniques (decision trees, artificial neural networks and support vector machines) along with the most commonly used statistical analysis technique logistic regression to develop prediction models for prostate cancer survivability. The data set contained around 120 000 records and 77 variables. A k-fold cross-validation methodology was used in model building, evaluation and comparison. The results showed that support vector machines are the most accurate predictor (with a test set accuracy of 92.85%) for this domain, followed by artificial neural networks and decision trees.

AB - Even though cancer research has traditionally been clinical and biological in nature, in recent years data driven analytic studies have become a common complement. In medical domains where data and analytics driven research is successfully applied, new and novel research directions are identified to further advance the clinical and biological studies. In this research, we used three popular data mining techniques (decision trees, artificial neural networks and support vector machines) along with the most commonly used statistical analysis technique logistic regression to develop prediction models for prostate cancer survivability. The data set contained around 120 000 records and 77 variables. A k-fold cross-validation methodology was used in model building, evaluation and comparison. The results showed that support vector machines are the most accurate predictor (with a test set accuracy of 92.85%) for this domain, followed by artificial neural networks and decision trees.

KW - Cancer

KW - Classification

KW - Data mining

KW - Decision trees

KW - Neural networks

KW - SEER

KW - Sensitivity analysis

UR - http://www.scopus.com/inward/record.url?scp=59349116912&partnerID=8YFLogxK

U2 - 10.1111/j.1468-0394.2008.00480.x

DO - 10.1111/j.1468-0394.2008.00480.x

M3 - Article

AN - SCOPUS:59349116912

VL - 26

SP - 100

EP - 112

JO - Expert Systems

JF - Expert Systems

SN - 0266-4720

IS - 1

ER -