gistfile1.r

train <- read.csv("~/code/informs/TrainingData.csv", row.names=1, header=TRUE)
test <- read.csv("~/code/informs/ResultData.csv", row.names=1, header=TRUE)

for (col in c("Variable142OPEN", "Variable142LOW", "Variable142HIGH",
              "Variable142LAST", "Variable158OPEN", "Variable158LOW",
              "Variable158HIGH", "Variable158LAST")) {
  train[col] <- NULL
  test[col] <- NULL
}
train[is.na(train)] <- 0.0
test[is.na(test)] <- 0.0

tot <- NULL
for (i in 1:10) {
  sam <- train[sample(1:nrow(train), replace=TRUE),]
  mod <- glm(TargetVariable ~ ., data=sam, family=binomial(link="logit"))
  if (is.null(tot)) {
    tot <- predict(mod, test, type="response")
  } else {
    tot <- tot + predict(mod, test, type="response")
  }
}
avg <- tot / 10
avg <- as.data.frame(avg)

avg <- avg + runif(nrow(avg), -0.05, 0.05)
avg[avg > 1] <- runif(length(avg[avg > 1]), 0.95, 1.0)
avg[avg < 0] <- runif(length(avg[avg < 0]), 0.0, 0.05)

write.csv(avg, "~/out.csv")