Wisconsin Breast Cancer Diagnostic Data Set
Vijayakumar Jawaharlal
May 3, 2014
This is the sample data set used in “Machine Learning with R” book to teach kNN algorithm.
Data exploration
library(caret)## Loading required package: lattice
## Loading required package: ggplot2setwd("/Users/njvijay/big_data/Github/MachineLearning-R/knn/data")
srcData <- read.csv("wisc_bc_data.csv")
str(srcData)## 'data.frame': 569 obs. of 32 variables:
## $ id : int 87139402 8910251 905520 868871 9012568 906539 925291 87880 862989 89827 ...
## $ diagnosis : Factor w/ 2 levels "B","M": 1 1 1 1 1 1 1 2 1 1 ...
## $ radius_mean : num 12.3 10.6 11 11.3 15.2 ...
## $ texture_mean : num 12.4 18.9 16.8 13.4 13.2 ...
## $ perimeter_mean : num 78.8 69.3 70.9 73 97.7 ...
## $ area_mean : num 464 346 373 385 712 ...
## $ smoothness_mean : num 0.1028 0.0969 0.1077 0.1164 0.0796 ...
## $ compactness_mean : num 0.0698 0.1147 0.078 0.1136 0.0693 ...
## $ concavity_mean : num 0.0399 0.0639 0.0305 0.0464 0.0339 ...
## $ points_mean : num 0.037 0.0264 0.0248 0.048 0.0266 ...
## $ symmetry_mean : num 0.196 0.192 0.171 0.177 0.172 ...
## $ dimension_mean : num 0.0595 0.0649 0.0634 0.0607 0.0554 ...
## $ radius_se : num 0.236 0.451 0.197 0.338 0.178 ...
## $ texture_se : num 0.666 1.197 1.387 1.343 0.412 ...
## $ perimeter_se : num 1.67 3.43 1.34 1.85 1.34 ...
## $ area_se : num 17.4 27.1 13.5 26.3 17.7 ...
## $ smoothness_se : num 0.00805 0.00747 0.00516 0.01127 0.00501 ...
## $ compactness_se : num 0.0118 0.03581 0.00936 0.03498 0.01485 ...
## $ concavity_se : num 0.0168 0.0335 0.0106 0.0219 0.0155 ...
## $ points_se : num 0.01241 0.01365 0.00748 0.01965 0.00915 ...
## $ symmetry_se : num 0.0192 0.035 0.0172 0.0158 0.0165 ...
## $ dimension_se : num 0.00225 0.00332 0.0022 0.00344 0.00177 ...
## $ radius_worst : num 13.5 11.9 12.4 11.9 16.2 ...
## $ texture_worst : num 15.6 22.9 26.4 15.8 15.7 ...
## $ perimeter_worst : num 87 78.3 79.9 76.5 104.5 ...
## $ area_worst : num 549 425 471 434 819 ...
## $ smoothness_worst : num 0.139 0.121 0.137 0.137 0.113 ...
## $ compactness_worst: num 0.127 0.252 0.148 0.182 0.174 ...
## $ concavity_worst : num 0.1242 0.1916 0.1067 0.0867 0.1362 ...
## $ points_worst : num 0.0939 0.0793 0.0743 0.0861 0.0818 ...
## $ symmetry_worst : num 0.283 0.294 0.3 0.21 0.249 ...
## $ dimension_worst : num 0.0677 0.0759 0.0788 0.0678 0.0677 ...#Removing id field which is not required for prediction.
wcData <- srcData[,-1]
table(wcData$diagnosis)##
## B M
## 357 212prop.table(table(wcData$diagnosis))##
## B M
## 0.6274 0.3726It has about 569 observations(biopsies) with 32 variables. diagnosis is the response variable with 2 levels B-begnin and M - melignant. All of the predictors are quantitative with different measurement. These predictors needs to be standardize before applying kNN algorithm.
Data Spliting
Reserve 20% data for validation. 80% for cross validation and model building
set.seed(456)
idx <- createDataPartition(wcData$diagnosis,p = 0.8,list = FALSE)
train <- wcData[idx,]
test <- wcData[-idx,]
#Checking proportion of data spliting
prop.table(table(train$diagnosis))##
## B M
## 0.6272 0.3728prop.table(table(test$diagnosis))##
## B M
## 0.6283 0.3717% is maintained in the data spliting
Setting parallelism
library(doMC)## Loading required package: foreach
## Loading required package: iterators
## Loading required package: parallelregisterDoMC(cores = 3)Building the model
set.seed(400)
ctrl <- trainControl(method="repeatedcv",repeats = 3,classProbs=TRUE,summaryFunction = twoClassSummary)
knnFit <- train(diagnosis ~ ., data = train, method = "knn", trControl = ctrl, preProcess = c("center","scale"), tuneLength = 31)## Loading required package: pROC
## Type 'citation("pROC")' for a citation.
##
## Attaching package: 'pROC'
##
## The following objects are masked from 'package:stats':
##
## cov, smooth, var## Warning: The metric "Accuracy" was not in the result set. ROC will be used
## instead.knnFit## k-Nearest Neighbors
##
## 456 samples
## 30 predictors
## 2 classes: 'B', 'M'
##
## Pre-processing: centered, scaled
## Resampling: Cross-Validated (10 fold, repeated 3 times)
##
## Summary of sample sizes: 411, 410, 410, 411, 411, 410, ...
##
## Resampling results across tuning parameters:
##
## k ROC Sens Spec ROC SD Sens SD Spec SD
## 5 1 1 0.9 0.02 0.02 0.06
## 7 1 1 0.9 0.02 0.02 0.07
## 9 1 1 0.9 0.01 0.01 0.06
## 10 1 1 0.9 0.01 0.01 0.06
## 10 1 1 0.9 0.01 0.007 0.07
## 20 1 1 0.9 0.01 0.02 0.07
## 20 1 1 0.9 0.01 0.01 0.07
## 20 1 1 0.9 0.01 0.009 0.07
## 20 1 1 0.9 0.01 0.009 0.07
## 20 1 1 0.9 0.01 0.01 0.07
## 20 1 1 0.9 0.01 0.01 0.07
## 30 1 1 0.9 0.01 0.01 0.07
## 30 1 1 0.9 0.01 0.01 0.07
## 30 1 1 0.9 0.01 0.01 0.07
## 30 1 1 0.9 0.01 0.01 0.07
## 40 1 1 0.9 0.01 0.01 0.07
## 40 1 1 0.9 0.01 0.01 0.07
## 40 1 1 0.9 0.01 0.01 0.07
## 40 1 1 0.9 0.01 0.01 0.07
## 40 1 1 0.9 0.01 0.01 0.08
## 40 1 1 0.9 0.008 0.01 0.07
## 50 1 1 0.9 0.008 0.01 0.07
## 50 1 1 0.9 0.008 0.01 0.07
## 50 1 1 0.9 0.008 0.01 0.07
## 50 1 1 0.9 0.008 0.01 0.07
## 60 1 1 0.9 0.008 0.01 0.07
## 60 1 1 0.9 0.008 0.01 0.07
## 60 1 1 0.8 0.008 0.01 0.08
## 60 1 1 0.8 0.009 0.01 0.09
## 60 1 1 0.8 0.009 0.01 0.09
## 60 1 1 0.8 0.009 0.01 0.09
##
## ROC was used to select the optimal model using the largest value.
## The final value used for the model was k = 45.plot(knnFit)
Validate the model on the test data
knnPredict <- predict(knnFit,newdata = test)
#Get the confusion matrix to see accuracy value and other parameter values
confusionMatrix(knnPredict, test$diagnosis )## Confusion Matrix and Statistics
##
## Reference
## Prediction B M
## B 71 4
## M 0 38
##
## Accuracy : 0.965
## 95% CI : (0.912, 0.99)
## No Information Rate : 0.628
## P-Value [Acc > NIR] : <2e-16
##
## Kappa : 0.923
## Mcnemar's Test P-Value : 0.134
##
## Sensitivity : 1.000
## Specificity : 0.905
## Pos Pred Value : 0.947
## Neg Pred Value : 1.000
## Prevalence : 0.628
## Detection Rate : 0.628
## Detection Prevalence : 0.664
## Balanced Accuracy : 0.952
##
## 'Positive' Class : B
## Measuring performance using ROCR package
knnperfPredict <- predict(knnFit,newdata = test,type="prob")
knnperfPredictB <- knnperfPredict$B
#Creating dummy vars
testdiagnosis <- ifelse(test$diagnosis == 'B', 1,0)
library(ROCR)## Loading required package: gplots
## KernSmooth 2.23 loaded
## Copyright M. P. Wand 1997-2009
##
## Attaching package: 'gplots'
##
## The following object is masked from 'package:stats':
##
## lowessperfprod <- prediction(knnperfPredictB, testdiagnosis)
perf <- performance(perfprod,"tpr","fpr")
plot(perf)
perf <- performance(perfprod,"sens","spec")
plot(perf)
ROCR supports only binary classification. It is a limitation.