# See if directory exists
dir.exists('FootballTemp')

[1] FALSE

# create it
dir.create('FootballTemp')
# check again
dir.exists('FootballTemp')

[1] TRUE

download.file('http://www.jaredlander.com/data/Football1415.tar.gz', 
              destfile='FootballTemp/football.tar.gz', 
              method='curl')

getXKCD('1168')

untar('FootballTemp/football.tar.gz', exdir='FootballFiles')

dir('FootballFiles')

[1] "pbp-2014.csv" "pbp-2015.csv"

unlink('FootballTemp/football.tar.gz')
dir('FootballTemp')

character(0)

file.info('FootballFiles/pbp-2014.csv')

                               size isdir mode               mtime
FootballFiles/pbp-2014.csv 10280324 FALSE  666 2016-03-25 00:14:23
                                         ctime               atime exe
FootballFiles/pbp-2014.csv 2016-04-04 22:48:47 2016-04-04 22:48:47  no

dir('FootballFiles') %>% file.info

             size isdir mode mtime ctime atime  exe
pbp-2014.csv   NA    NA <NA>  <NA>  <NA>  <NA> <NA>
pbp-2015.csv   NA    NA <NA>  <NA>  <NA>  <NA> <NA>

dir('FootballFiles', full.names=TRUE) %>% file.info

                               size isdir mode               mtime
FootballFiles/pbp-2014.csv 10280324 FALSE  666 2016-03-25 00:14:23
FootballFiles/pbp-2015.csv 10671016 FALSE  666 2016-03-25 00:14:23
                                         ctime               atime exe
FootballFiles/pbp-2014.csv 2016-04-04 22:48:47 2016-04-04 22:48:47  no
FootballFiles/pbp-2015.csv 2016-04-04 22:48:47 2016-04-04 22:48:47  no

file.rename(from=dir('FootballFiles', full.names=TRUE), 
            to=sprintf('FootballFiles/Football%s.csv', 14:15))

[1] TRUE TRUE

dir('FootballFiles')

[1] "Football14.csv" "Football15.csv"

dir.create('FootballFiles/Backup')
file.copy(dir('FootballFiles', full.names=TRUE, pattern='\\.csv'), 
          sprintf('FootballFiles/Backup/Footballl%s.csv', 14:15))

[1] TRUE TRUE

dir('FootballFiles', recursive=TRUE)

[1] "Backup/Footballl14.csv" "Backup/Footballl15.csv" "Football14.csv"        
[4] "Football15.csv"        

count.fields('FootballFiles/Football14.csv', sep=',') %>% head(15)

 [1] 45 45 45 45 15 45 45 45 45 45 45 45 45 45 45

count.fields('FootballFiles/Football15.csv', sep=',') %>% head(15)

 [1] 45 45 45 45 45 45 45 45 45 NA 15 45 45 45 45

system('wc -l FootballFiles/Football14.csv')

45696 FootballFiles/Football14.csv

system('wc -l FootballFiles/Football15.csv')

46278 FootballFiles/Football15.csv

dataPath <- 'FootballFiles'
file.path(dataPath, 'Football14.csv')

[1] "FootballFiles/Football14.csv"

file.path(dataPath, 'Football15.csv')

[1] "FootballFiles/Football15.csv"

theFiles <- dir(dataPath, pattern='\\.csv', full.names=TRUE)
games <- theFiles %>% map_df(read.csv2, sep=',', header=TRUE, stringsAsFactors=FALSE)

DT::datatable(data=games%>% slice(sample(nrow(games), size=500, replace=FALSE)), 
              rownames=FALSE,
              options = list(
                  dom = "rtiS",
                  scrollY = 400, scrollX=TRUE,
                  scrollCollapse = TRUE),
              filter=list(position='top')
)

oneOff <- games %>%
    filter(OffenseTeam == 'NYG', PlayType %in% c('PASS', 'RUSH')) %>%
    mutate(PlayType=factor(PlayType, levels=c('RUSH', 'PASS')), 
           Down=factor(Down, levels=c(1, 2, 3, 4)))

passRushMod <- glm(PlayType ~ Down + ToGo - 1, data=oneOff, family=binomial)
coefplot(passRushMod, trans=arm::invlogit, title='Probability of Pass')

# make grid of scenarios
scenarios <- expand.grid(ToGo=1:15, Down=1:4) %>% as.tbl %>% 
    mutate(Down=factor(Down, levels=c(1, 2, 3, 4)))
# make prediction based on model
scenarioPredict <- predict(passRushMod, 
                           newdata=scenarios, type='response', se.fit=TRUE)
# build confidence intervals
scenarios <- scenarios %>% mutate(Prediction=scenarioPredict$fit, 
                                  Lower=Prediction - 2*scenarioPredict$se.fit,
                                  Upper=Prediction + 2*scenarioPredict$se.fit)

knitr::kable(head(scenarios))

ggplot(scenarios, aes(x=ToGo)) + scale_y_continuous(label=scales::percent) +
    geom_ribbon(aes(ymin=Lower, ymax=Upper), fill='lightgrey') +
    geom_line(aes(y=Prediction)) + facet_wrap(~Down, nrow=2)

eliPage <- read_html('http://www.pro-football-reference.com/players/M/MannEl00.htm')

eliStats <- eliPage %>% html_nodes("#passing") %>% 
    html_table(header=TRUE) %>% getElement(1)
useful::topleft(eliStats, c=7, r=8)

   Year Age  Tm Pos No.  G GS
1  2004  23 NYG  qb  10  9  7
2  2005  24 NYG  QB  10 16 16
3  2006  25 NYG  QB  10 16 16
4  2007  26 NYG  QB  10 16 16
5 2008*  27 NYG  QB  10 16 16
6  2009  28 NYG  QB  10 16 16
7  2010  29 NYG  QB  10 16 16
8 2011*  30 NYG  QB  10 16 16

dir.create('results')
ggsave('results/EliPass.png')
write.table(eliStats, file='results/eliStats.csv', sep=',', row.names=FALSE)

[1] TRUE

[1] TRUE

repo <- repository(getwd())
add(repo, file.path('results', c('eliPass.png', 'eliStats.csv')))
commit(repo, message='Tracking plot and csv')
push(repo)

footballResults <- mime(
    To = "jared@landeranalytics.com",
    From = "jared@jaredlander.com",
    Subject = "Eli Results",
    body = "See the attached graph and data.") %>% 
    attach_file('results/EliPass.png') %>% 
    attach_file('results/eliStats.csv')
send_message(footballResults)

• Chief Data Scientist of Lander Analytics
• Author of R for Everyone
  
• Adjunct Professor at Columbia University
  
• Organizer of New York Open Statistical Programming (The R) Meetup
  
• Website: http://www.jaredlander.com

• rvest
• ggplot2
• dplyr
• purrr
• coefplot
• magritrr
• useful

• R
• RStudio
• knitr
• Pandoc
• ioslides