# Creates a simple random forest benchmark

library(randomForest)
library(readr)
library(caret)

set.seed(0)

input <- read_csv("../input/train.csv")
test <- read_csv("../input/test.csv")

train <- createDataPartition(y=input$label, p=0.5,times=1, list=FALSE)
numTrain <- input[train,]
numTest <- input[-train,]

varNames <- names(numTrain)
varNames <- varNames[!varNames %in% c("label")]
varNames1 <- paste(varNames, collapse = "+")
labels <- as.factor(numTrain$label)
form <- as.formula(paste("labels", varNames1, sep = " ~ "))

rf <- randomForest(form, data=numTrain, ntree=50)
prediction <- predict(rf, type="class", newdata=test)

predictions <- data.frame(ImageId=1:nrow(test), Label=prediction)

write_csv(predictions, "sol1.csv")