Kaggle Bike Sharing Demand Competition - Linear Regression Model - R

kaggle_bikesharing_1.R

library(caTools)
library(caret)
library(Amelia)

data.dir <- "/Documents/Data\ Science/Kaggle/Bike\ Sharing\ Demand/"

train.file <- paste0(data.dir, "train.csv")
test.file <- paste0(data.dir, "test.csv")

bk <- read.csv(train.file)
test <- read.csv(test.file)


#################################################
#Descriptive statistics and cleaning the dataset
################################################

str(bk)
summary(bk)
str(test)
summary(test)


bk['casual'] <- NULL
bk['registered'] <- NULL

table(complete.cases(bk)) #all cases are complete

#Imputation
#impute <- amelia(x = bk, m = 1, noms = c('holiday', 'workingday', 'weather'), idvars = c('datetime')) #set datetime as an id variable so removed from model
#plot(impute)
#bk <- as.data.frame(impute$imputations[1])

str(bk)

#splitting the data set



#copying the dataset
bk.copy <- bk
test.copy <- test

#deleting incomplete cases 
#bk <- na.omit(bk)

#deleting date rows
bk[,1]<- NULL
test[,1] <- NULL


#dv is count

#splitting into testing and training set

bk.split <- sample.split(bk$count, SplitRatio = 0.7)
bk.train <- subset(bk, bk.split == TRUE)
bk.test <- subset(bk, bk.split == FALSE)

#Checking the split
str(bk.train)
str(bk.test)
summary(bk.train)
summary(bk.test)

##############
#Models
##############

#Linear Model -- Baseline
lm1 <- lm(count~., data = bk.train)
summary(lm1)
lm1.test.predict <- predict(lm1, newdata = bk.test)



################
#Predictions for testing set
###############

table(complete.cases(test))
#all cases are complete

lm1.kaggle.predict <- predict(lm1, newdata = test)
str(lm1.kaggle.predict)

final.results <- data.frame(lm1.kaggle.predict, test.copy[,1])

final.results.copy <- final.results

final.results[,1] <- final.results.copy[,2]
final.results[,2] <- final.results.copy[,1]


colnames(final.results)[1] <- 'datetime'
colnames(final.results)[2] <- 'count'

which(final.results$count < 0)
which(is.na(final.results$count))
final.results[which(final.results$count < 0),]$count <- 0
#eliminating the negative variables

write.table(final.results, file = paste0(data.dir,'results.csv'), sep = ",", row.names = FALSE, quote = FALSE)

Thanks for this. I was having trouble outputting my data cause I'm new to R.

No problem! Glad it was helpful