支持率と勝率の関係性を調べる

winper_and_support.R

#参考
#http://www.amstat.org/chapters/boston/nessis07/presentation_material/Victor_Lo.pdf
#http://hub.hku.hk/bitstream/10722/60987/4/Content_11.pdf?accept=1

ws <- read.csv("winper_and_support.csv", header=F)

ws <- ws[which(ws$V1 != 0),]

ws <- ws[which(ws$V3 > 100),]

x <- matrix(0, ncol=2, nrow=nrow(ws)*(nrow(ws)-1))
k <- 1
for (i in 1:nrow(ws)) {
  for (j in 1:nrow(ws)) {
    if (i != j) {
      print(k)
      x[k,1] <-
        log(ws[i,]$V1/ws[j,]$V1)
      x[k,2] <-
        log(ws[i,]$V2/ws[j,]$V2)
      k <- k + 1
    }
  }
}

x <- as.data.frame(x)
names(x) <- c("V1", "V2")

model.lm <- lm(V1 ~ V2 - 1, x)

summary(model.lm)

#
#Call:
#lm(formula = V1 ~ V2 - 1, data = x)
#
#Residuals:
#    Min      1Q  Median      3Q     Max 
#-3.0004 -0.3485  0.0000  0.3485  3.0004 
#
#Coefficients:
#    Estimate Std. Error t value Pr(>|t|)    
#V2 0.9538157  0.0005819    1639   <2e-16 ***
#---
#Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#
#Residual standard error: 0.5816 on 636005 degrees of freedom
#Multiple R-squared:  0.8086,    Adjusted R-squared:  0.8086 
#F-statistic: 2.687e+06 on 1 and 636005 DF,  p-value: < 2.2e-16

#βが1を下回るから何か根本的に勘違いしているかも…？

winper_and_support.sql

.separator ,
.output winper_and_support.csv
select
  avg(order_of_finish = 1),
  0.788 / (odds - 0.1),
  count(*) 
from 
  (select cast(order_of_finish as int) order_of_finish, odds from race_result) 
where 
  order_of_finish <> 0 
group by 
  odds;