Module 3: Exercise Results

data(BreastCancer)
bc <- BreastCancer
for (k in 2:10) # altered for current lab
    bc[,k] <- as.numeric(bc[,k]) 
head(bc)

##        Id Cl.thickness Cell.size Cell.shape Marg.adhesion Epith.c.size Bare.nuclei Bl.cromatin
## 1 1000025            5         1          1             1            2           1           3
## 2 1002945            5         4          4             5            7          10           3
## 3 1015425            3         1          1             1            2           2           3
## 4 1016277            6         8          8             1            3           4           3
## 5 1017023            4         1          1             3            2           1           3
## 6 1017122            8        10         10             8            7          10           9
##   Normal.nucleoli Mitoses     Class
## 1               1       1    benign
## 2               2       1    benign
## 3               1       1    benign
## 4               7       1    benign
## 5               1       1    benign
## 6               7       1 malignant

Explore missingness:

suppressMessages(require(plotrix))

#' show data missingness as a chequered matrix
#' 
#' @param x (matrix) data matrix.
#' @param outFile (char) path to file for printing graph
#' @param wd (numeric) width in inches
#' @param ht (numeric) height in inches
#' @return plots missingness matrix to file
#' @import plotrix
#' @export
plotMissMat <- function(x,xlab="columns",
        ylab="rows",border=NA) {
    
    x <- !is.na(x)
    class(x) <- "numeric"
    color2D.matplot(x,show.values=FALSE,axes=FALSE,
        cs1=c(1,0),cs2=c(1,0),cs3=c(1,0),border=border,
        cex=0.8,
        xlab=xlab,ylab=ylab)
}

Explore missingness:

plotMissMat(bc)

colSums(is.na(bc))

##              Id    Cl.thickness       Cell.size      Cell.shape   Marg.adhesion    Epith.c.size 
##               0               0               0               0               0               0 
##     Bare.nuclei     Bl.cromatin Normal.nucleoli         Mitoses           Class 
##              16               0               0               0               0

Plot relationship between variables

require(ggplot2)

ggplot(bc,aes(Cell.size,Normal.nucleoli)) + geom_point()

ggplot(bc) + geom_boxplot(aes(factor(Class), Normal.nucleoli))

ggplot(bc) + geom_boxplot(aes(factor(Class), Cl.thickness))

ggplot(bc) + geom_boxplot(aes(factor(Class), Bare.nuclei))

## Warning: Removed 16 rows containing non-finite values (`stat_boxplot()`).

Fit a binary outcome model:

mod <- glm(
    Class ~ Cl.thickness + Bare.nuclei + Normal.nucleoli + Mitoses + Bl.cromatin, 
    bc,
    family="binomial")
summary(mod)

## 
## Call:
## glm(formula = Class ~ Cl.thickness + Bare.nuclei + Normal.nucleoli + 
##     Mitoses + Bl.cromatin, family = "binomial", data = bc)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -3.7840  -0.1459  -0.0768   0.0324   2.7798  
## 
## Coefficients:
##                 Estimate Std. Error z value Pr(>|z|)    
## (Intercept)     -9.69635    0.98763  -9.818  < 2e-16 ***
## Cl.thickness     0.63413    0.12267   5.169 2.35e-07 ***
## Bare.nuclei      0.50988    0.08228   6.197 5.76e-10 ***
## Normal.nucleoli  0.35765    0.10040   3.562 0.000368 ***
## Mitoses          0.52677    0.27923   1.887 0.059223 .  
## Bl.cromatin      0.61398    0.14225   4.316 1.59e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 884.35  on 682  degrees of freedom
## Residual deviance: 120.81  on 677  degrees of freedom
##   (16 observations deleted due to missingness)
## AIC: 132.81
## 
## Number of Fisher Scoring iterations: 8