2011 NHL D Statistics

NHL 2011 Defensive Stats.R

# set working directory
setwd("/My Dropbox/Projects/NHL Defensemen 2011 performance")

# load the libraries I commonly use
library(XML)
library(plyr)
library(lubridate)
library(ggplot2)

# grab the data
URL <- "http://www.hockey-reference.com/leagues/NHL_2011_skaters.html"
tables <- readHTMLTable(URL)$stats
head(tables)

# filter on D
ds <- tables[tables$Pos == 'D', ]
nrow(ds)  # number of records

# change data types -- probably an easier way, but this helped me learn R
str(ds)
for (i in c(1,3,6:19)) {
  ds[,i] <- as.numeric(as.character(ds[,i])) # important! -- convert factor to string first
}
for (i in c(2, 4:5, 20)) {
  ds[,i] <- as.character(ds[,i])
}

# lets cut on games played to "core" set of players -- upper 50%
summary(ds$GP)
hist(ds$GP, xlab="Games played", main="Distribution of games played")
ds <- ds[ds$GP >= median(ds$GP),]

# lets look at the distribution of +/-
names(ds)[11] <- "plusmin"
summary(ds$plusmin)
hist(ds$plusmin)

# sort the dataframe
sorted.pm <- ds[order(ds$plusmin, decreasing=T), ]

# top 25
head(sorted.pm, n=25)

# plot plusmin and time on ice
plot(ds$TOI, ds$plusmin, xlab="+/-", ylab="Points", pch=20, cex=.8)

# sort the dataframe on TOI
sorted.toi <- ds[order(ds$TOI, decreasing=T), ]

# top 25
head(sorted.toi, n=25)

# Zee was top on +/- and top 3 in TOI..... +/- not the best stat, but coupled with TOI, its a start IMO