Bike Sharing Demand

library(data.table)
train<-fread(file="C:\\Users\\r631758\\Desktop\\r631758\\R codes\\BikeShare\\train.csv",stringsAsFactors = FALSE)
test<-fread(file="C:\\Users\\r631758\\Desktop\\r631758\\R codes\\BikeShare\\test.csv",stringsAsFactors = FALSE )
test$registered=0
test$casual=0
test$count=0
data=rbind(train,test)

summary(data)

   datetime             season         holiday          workingday        weather           temp           atemp          humidity        windspeed          casual      
 Length:17379       Min.   :1.000   Min.   :0.00000   Min.   :0.0000   Min.   :1.000   Min.   : 0.82   Min.   : 0.00   Min.   :  0.00   Min.   : 0.000   Min.   :  0.00  
 Class :character   1st Qu.:2.000   1st Qu.:0.00000   1st Qu.:0.0000   1st Qu.:1.000   1st Qu.:13.94   1st Qu.:16.66   1st Qu.: 48.00   1st Qu.: 7.002   1st Qu.:  0.00  
 Mode  :character   Median :3.000   Median :0.00000   Median :1.0000   Median :1.000   Median :20.50   Median :24.24   Median : 63.00   Median :12.998   Median :  3.00  
                    Mean   :2.502   Mean   :0.02877   Mean   :0.6827   Mean   :1.425   Mean   :20.38   Mean   :23.79   Mean   : 62.72   Mean   :12.737   Mean   : 22.56  
                    3rd Qu.:3.000   3rd Qu.:0.00000   3rd Qu.:1.0000   3rd Qu.:2.000   3rd Qu.:27.06   3rd Qu.:31.06   3rd Qu.: 78.00   3rd Qu.:16.998   3rd Qu.: 26.00  
                    Max.   :4.000   Max.   :1.00000   Max.   :1.0000   Max.   :4.000   Max.   :41.00   Max.   :50.00   Max.   :100.00   Max.   :56.997   Max.   :367.00  
   registered         count    
 Min.   :  0.00   Min.   :  0  
 1st Qu.:  0.00   1st Qu.:  0  
 Median : 23.00   Median : 28  
 Mean   : 97.44   Mean   :120  
 3rd Qu.:155.00   3rd Qu.:192  
 Max.   :886.00   Max.   :977  

par(mfrow=c(4,2))
par(mar=rep(2,4))
hist(data$season)
hist(data$weather)

hist(data$humidity)
hist(data$holiday)

hist(data$workingday)
hist(data$temp)

hist(data$atemp)
hist(data$windspeed)

prop.table(table(data$weather))

           1            2            3            4 
0.6567121238 0.2614649865 0.0816502676 0.0001726221 

change to factor

data$season=as.factor(data$season)
data$weather=as.factor(data$weather)
data$holiday=as.factor(data$holiday)
data$workingday=as.factor(data$workingday)

get time from date

data$hour=substr(data$datetime,12,13)
data$hour=as.factor(data$hour)

hour trend

train=data[as.integer(substr(data$datetime,9,10))<20, ]
test=data[as.integer(substr(data$data$datetime,9,10))>19, ]
boxplot(train$count~train$hour,xlab="hour",ylab="count of users")

boxplot(train$casual~train$hour,xlab="hour",ylab="count of users")

boxplot(train$registered~train$hour,xlab="hour",ylab="count of users")

log transformation of count

boxplot(log(train$count)~train$hour,xlab="hour",ylab="log(count)")

day trend

date<-substr(data$datetime,1,10)
days<-weekdays(as.Date(date))
data$day=days
train=data[as.integer(substr(data$datetime,9,10))<20, ]
test=data[as.integer(substr(data$data$datetime,9,10))>19, ]
boxplot(train$casual~train$day,xlab="hour",ylab="count of casual users")

boxplot(train$registered~train$day,xlab="hour",ylab="count of registered users")

year trend

data$year=substr(data$datetime,1,4)
data$year=as.factor(data$year)
train=data[as.integer(substr(data$datetime,9,10))<20,]
test=data[as.integer(substr(data$datetime,9,10))>19,]
boxplot(train$count~train$year,xlab="year", ylab="count")

weather

boxplot(train$casual~train$weather,xlab="weather",ylab="count of casual users")

boxplot(train$registered~train$weather,xlab="weather",ylab="count of registered users")

for continuous variables

sub=data.frame(train$registered, train$casual, train$count, train$temp, train$humidity, train$atemp, train$windspeed)
cor(sub)

                 train.registered train.casual train.count  train.temp train.humidity train.atemp train.windspeed
train.registered       1.00000000   0.49724969   0.9709481  0.31857128    -0.26545787  0.31463539      0.09105166
train.casual           0.49724969   1.00000000   0.6904136  0.46709706    -0.34818690  0.46206654      0.09227619
train.count            0.97094811   0.69041357   1.0000000  0.39445364    -0.31737148  0.38978444      0.10136947
train.temp             0.31857128   0.46709706   0.3944536  1.00000000    -0.06494877  0.98494811     -0.01785201
train.humidity        -0.26545787  -0.34818690  -0.3173715 -0.06494877     1.00000000 -0.04353571     -0.31860699
train.atemp            0.31463539   0.46206654   0.3897844  0.98494811    -0.04353571  1.00000000     -0.05747300
train.windspeed        0.09105166   0.09227619   0.1013695 -0.01785201    -0.31860699 -0.05747300      1.00000000

summary(train$hour)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
   1.00    7.00   13.00   12.54   19.00   24.00 

use rpart for decision tree

library(rpart)
library(rattle)

package <U+393C><U+3E31>rattle<U+393C><U+3E32> was built under R version 3.4.3Error in inDL(x, as.logical(local), as.logical(now), ...) : 
  unable to load shared object 'C:/Users/r631758/Documents/R/win-library/3.4/RGtk2/libs/x64/RGtk2.dll':
  LoadLibrary failure:  The specified module could not be found.

Failed to load RGtk2 dynamic library, attempting to install it.trying URL 'http://ftp.gnome.org/pub/gnome/binaries/win64/gtk+/2.22/gtk+-bundle_2.22.1-20101229_win64.zip'
Content type 'application/zip' length 25830230 bytes (24.6 MB)
downloaded 24.6 MB

'C:\Users\r631758\Documents\R\win-library\3.4\RGtk2\gtk\x64\etc\gtk-2.0' already existsLearn more about GTK+ at http://www.gtk.org
If the package still does not load, please ensure that GTK+ is installed and that it is on your PATH environment variable
IN ANY CASE, RESTART R BEFORE TRYING TO LOAD THE PACKAGE AGAIN
Rattle: A free graphical interface for data science with R.
Version 5.1.0 Copyright (c) 2006-2017 Togaware Pty Ltd.
Type 'rattle()' to shake, rattle, and roll your data.

library(rpart.plot)
library(RColorBrewer)
d=rpart(registered~hour,data=train)
fancyRpartPlot(d)

data$year_part[data$year=='2011']=1
data$year_part[data$year=='2011' & data$month>3]=2
data$year_part[data$year=='2011' & data$month>6]=3
data$year_part[data$year=='2011' & data$month>9]=4
data$year_part[data$year=='2012']=5
data$year_part[data$year=='2012' & data$month>3]=6
data$year_part[data$year=='2012' & data$month>6]=7
data$year_part[data$year=='2012' & data$month>9]=8
table(data$year_part)

   1    2    3    4    5    6    7    8 
2067 2183 2192 2203 2176 2182 2208 2168 

model building

library(randomForest)
data$hour=as.factor(data$hour)
data$day_type=as.factor(data$day_type)
data$day=as.factor(data$day)
data$logreg=log(data$registered+1)
data$logcas=log(data$casual+1)
train=data[as.integer(substr(data$datetime,9,10))<20,]
test=data[as.integer(substr(data$datetime,9,10))>19,]
#predicting the log of registered users.
set.seed(415)
fit1 <- randomForest(logreg ~ hour +workingday+day+holiday+ day_type +humidity+atemp+windspeed+season+weather+dp_reg+weekend+year+year_part, data=train,importance=TRUE, ntree=250)