using RDatasets
using Gadfly
using DataFrames


iris = dataset("datasets", "iris")

pwd()

cd("C:\\Users\\dell\\Desktop")

pwd()

readdir()


df=readtable("adult.data.txt",header=false)

size(df)

head(df)

plot(df,x="x1" ,color="x15",Geom.histogram)

describe(df)


plot(iris,x="SepalLength",y="SepalWidth",color="Species")

describe(df[:x10]) #refer to parts of a DataFrame

describe(df[:x1])

test1=(23,56,78,89)

mean(test1)

test3=(NA,34,67,89)


mean(removeNA(test3))

using DataArrays
test2=@data([23,56,78,89,NA]) #Only DataArrays can have missing values

mean(test2)

mean(removeNA(test2))

#Note removeNA has been changed to dropna in Julia 0.3

?removeNA #Notice the help

plot(df,y="x1",Geom.boxplot)

using PyPlot

boxplot(df[:x1])

pwd()