bikemember:

install.packages("XML")

library(dplyr)
library(zoo)
library(lubridate)
library(XML)
tripdata <- read.csv("R/input/201805-capitalbikeshare-tripdata.csv")
View(tripdata)
colnames(tripdata)
tripdata <- tripdata[!(is.na(tripdata$End.station) | tripdata$End.station==""), ]

# External data
xmlfile <- xmlTreeParse('R/input/stations.xml',useInternalNodes = TRUE)
topxml <- xmlRoot(xmlfile)
topxml <- xmlSApply(topxml, function(x) xmlSApply(x, xmlValue))
station_data <- data.frame(t(topxml), row.names=NULL)
station_data$terminalName <- as.numeric(station_data$terminalName)
station_data$lat <- as.numeric(station_data$lat)
station_data$long <- as.numeric(station_data$long)
station_data$name <- as.character(station_data$name)
View(station_data)
write.csv(station_data,file = "R/input/station_data.csv")

temp <- tripdata
# write.csv(tripdata,file = "df1.csv")
# View(temp)
# View(station_data)
# View(tripdata)
# glimpse(station_data)
# temp <- cbind("ss_id","ss_terminalName","ss_lat","ss_long","ss_nbBikes","ss_nbEmptyDocks")
# glimpse(temp)


temp$name <- temp$Start.station
dim_desc(tripdata)
df <- merge(temp,station_data,by="name",all.x = TRUE)
df$name <- df$End.station
df1 <- merge(df,station_data,by="name",all.x = TRUE,suffixes ="es")
dim_desc(df1)
write.csv(df1,file = "R/input/df1.csv")
colnames(df1)

cols <- c("Member.type","Duration","terminalNamees","lates","longes","nbBikeses",
          "nbEmptyDockses","terminalNameNA","latNA","longNA","nbBikesNA","nbEmptyDocksNA")

df.fl <- df1[,c(cols)]
dim_desc(df.fl)

install.packages('caTools')
library(caTools)
split <- sample.split(df.fl$Member.type, SplitRatio = 0.75)
split
#get training and test data
train <- subset(df.fl, split == TRUE)
test <- subset(df.fl, split == FALSE)

train[is.na(train)] <- 0.00001
dim_desc(train)
dim_desc(test)


#logistic regression model
model <- glm (Member.type ~ ., data = train, family = binomial)
summary(model)
predictions <- predict(model,test[,-1],type="response")


x = train[,-1]
y= train[,1]
sum(is.na(train))
sum(is.na(y))
x[is.na(x)] <- 0


#Fitting the Naive Bayes model
library(e1071)
nbmodel = train(x,y,'nb',trControl=trainControl(method='cv',number=2))
nbmodel
predict(nbmodel$finalModel,x[50,])
table(predict(nbmodel$finalModel,x[1:50,])$class,y[1:50])