Western Bulldogs - Player Stats

As I mentioned in my first post, my wife and joined the Western Bulldogs this year.  We received our first copy of "The Bulldog".  The player listing caught my eye :  stats about the players : age, weight, height, AFL Debut Year.

So here are my statistics and charts about this year's WB players:

- 47 players listed

- 6 players share birthdays with one other person

A plot of height versus weight shows a clear linear relationship:

 

 

From this relationship, we can find equations to predict height from weight, or vice versa:

          Height =  117.108 + 0.8078 * weight

          Weight = - 68.7018 + 0.8316 * height.

The R code to produce the chart above is :

library(ggplot2)

ggplot(wb, aes(WT,HT))+ geom_point() + scale_x_continuous("Weight") + scale_y_continuous("Height") + opts(title = "Western Bulldogs 2012 - Height * Weight") + geom_smooth(method = "lm")

This code produces a straight line smoother called "linear model".

The following code produces the same chart, with a different smoother:

 

 

ggplot(wb, aes(WT,HT))+ geom_point() + scale_x_continuous("Weight") + scale_y_continuous("Height") + opts(title = "Western Bulldogs 2012 - Height * Weight") + geom_smooth()

The geom_smooth() function in this case uses the default smoother "Loess" (for datasets up to n = 1,000 ). For the mathematically inclined, this smoother fits a polynomial line.

There is virtually no relationship between height and age:

  

ggplot(wb, aes(Age1Jan2012, HT))+ geom_point() + scale_x_continuous("Age") + scale_y_continuous("Height") + opts(title = "Western Bulldogs 2012 - Height * Age") + geom_smooth(method = "lm")

or

weight and age :

 

 

 

ggplot(wb, aes(Age1Jan2012, WT))+ geom_point() + scale_x_continuous("Age") + scale_y_continuous("Weight") + opts(title = "Western Bulldogs 2012 - Weight * Age") + geom_smooth(method = "lm")

And finally, here is the age distribution:

  

qplot(Age2, data = wb, geom = "histogram", binwidth = 1)+ scale_x_continuous("Age") + scale_y_continuous("Count") + opts(title = "Western Bulldogs 2012 - Number By Age")

 

 

 

 

Contested Posessions Plot Using ggplot2

Of interest to the R user (but unfortunately, not to the follower of footy ) is how to reproduce the Contested Possessions box-plot using ggplot2. Now ggplot2 is a well known package in R - ggplot2 is a powerful and flexible system for creating chart graphics - it's promise is extra flexibility and power than one gets with the charting functions in Base R.

ggplot2 has two options:

- qplot - this is the quick plotting function

- ggplot2 - this is the full plotting function, allowing almost unlimited customisation

My aim is to produce the box-plots initially using qplot, and then to recreate the plot in ggplot2.

This code produces the box-plot using qplot:

byafl <- with(afl, reorder(V2,V1))

qplot(byafl,V3, data = afl, geom = "boxplot", xlab = "Teams - Ladder Order", ylab = "Contested Possessions Per Game", main = "Contested Posessions 2011")

The equivalent code using ggplot2 is:

ggplot(afl, aes(byafl,V3))+ geom_boxplot() + scale_x_discrete("Teams - Ladder Order") + scale_y_continuous("Contested Possessions Per game") + opts(title = "Contested Possessions 2011")

Here is the plot using ggplo2 - is there any difference ?