Exploratory Data Analysis

Test of Association for Continuous Variables

This is an example using a correlation plot across all the variables in the dataset. Use patterns to understand the level of correlation between these variables.

corrgram(mtcars, order=NULL, lower.panel=panel.shade,
  upper.panel=NULL, text.panel=panel.txt,
  main="Car Milage Data (unsorted)")

Test of Association for Categorical Variables

Just like correlation for continuous variables we can use the following chi square plot to find the association between categorical variables using chi square test of associations and plotting it.

mushrooms$freq=1
mushrooms$veil.type = NULL
mat = matrix(nrow=23,ncol=23, byrow = TRUE, dimnames= list(names(mushrooms),names(mushrooms)))

for(i in 1:ncol(mushrooms))
{
  for(j in 1:ncol(mushrooms))
  {
    #print(i)
    #print(j)
    temp = xtabs(freq~mushrooms[,j]+mushrooms[,i], data = mushrooms)
    ##temp2= chisq.test(temp)
    mat[i,j]= cramersV(temp)
  }
}

## Error in chisq.test(...): 'x' must at least have 2 elements

cellcol<-color.scale(cbind(mat,c(-1,rep(1,22))),c(0,1),0,c(1,0))[,1:13]
par(mar = c(8,8,0.1,0.1) + 0.3)

color2D.matplot(mat,cellcolors=cellcol,show.legend=TRUE,show.values=1,
                axes=FALSE, xlab="",ylab="")

axis(1,at=0.5:22.5,las=2,labels=colnames(mat))
axis(2,at=0.5:22.5,las=2,labels=rev(rownames(mat)))

An Example of creating small multiples by different categorical variables to analyze the data

plotHist <- function(data_in, i) {
  data <- data.frame(x=data_in[[i]])
  p <- ggplot(data=data, aes(x=factor(x))) + stat_count() + xlab(colnames(data_in)[i]) + theme_light() + 
    theme(axis.text.x = element_text(angle = 90, hjust =1))
  return (p)
}
doPlots <- function(data_in, fun, ii, ncol=3) {
  pp <- list()
  for (i in ii) {
    p <- fun(data_in=data_in, i=i)
    pp <- c(pp, list(p))
  }
  do.call("grid.arrange", c(pp, ncol=ncol))
}

doPlots(mushrooms, fun = plotHist, ii = 16:23, ncol = 2)