## Importing packages

# This R environment comes with all of CRAN and many other helpful packages preinstalled.
# You can see which packages are installed by checking out the kaggle/rstats docker image:
# https://github.com/kaggle/docker-rstats

library(tidyverse) # metapackage with lots of helpful functions

## Running code

# In a notebook, you can run a single code cell by clicking in the cell and then hitting
# the blue arrow to the left, or by clicking in the cell and pressing Shift+Enter. In a script,
# you can run code by highlighting the code you want to run and then clicking the blue arrow
# at the bottom of this window.

## Reading in files

# You can access files from datasets you've added to this kernel in the "../input/" directory.
# You can see the files added to this kernel by running the code below.

empset <- read.csv("../input/MFG10YearTerminationData.csv")
## Saving data

# If you save any files or images, these will be put in the "output" directory. You
# can see the output directory by committing and running your kernel (using the
# Commit & Run button) and then checking out the compiled version of your kernel.

#setwd("D:/Kriti/practice/employee-attrition-data")
#empset<-read.csv("D:/Kriti/practice/employee-attrition-data/MFG10YearTerminationData.csv")

str(empset)
#View(empset)

summary(empset)

library(tidyr)
library(ggplot2)
library(dplyr)
library(plyr)

statusCount<-as.data.frame.matrix(empset %>%
                       group_by(STATUS_YEAR) %>%
                       select(STATUS) %>%
                       table())

statusCount$Total<-statusCount$ACTIVE+statusCount$TERMINATED

statusCount$PercentTerminate<-statusCount$TERMINATED/(statusCount$Total)*100


statusCount

mean(statusCount$PercentTerminate)


ggplot(empset)+geom_bar(aes(x=BUSINESS_UNIT, fill=STATUS))

TerminateData<- empset %>% filter(STATUS=="TERMINATED")

ggplot(TerminateData)+geom_bar(aes(x=STATUS_YEAR, fill=termtype_desc))

ggplot(TerminateData)+geom_bar(aes(x=STATUS_YEAR, fill=termreason_desc))

ggplot(TerminateData)+geom_bar(aes(x=as.factor(department_name), fill=as.factor(termreason_desc)))+
                               theme(axis.text.x = element_text(angle=90, hjust=1,vjust=0.5))


library(caret)

featurePlot(x=empset[,6:7],y=empset$STATUS,plot="density",auto.key=list(columns=2))

featurePlot(x=empset[,6:7],y=empset$STATUS,plot="box",auto.key=list(columns=2))

library(rattle)
library(magrittr)
library(randomForest)
crv$seed=42

set.seed(crv$seed)

empNum<-nrow(empset)

empTrain<-subset(empset,STATUS_YEAR<2015)
empTest<-subset(empset,STATUS_YEAR==2015)


MYinput= c("age", "length_of_service",    "gender_full", "STATUS_YEAR", "BUSINESS_UNIT")
MYnumeric= c("age", "length_of_service", "STATUS_YEAR")
MYcategoric = c("gender_full", "BUSINESS_UNIT")
MYtarget= "STATUS"
MYident = "EmployeeID"
MYTrainingData<-empTrain[c(MYinput, MYtarget)]
MYTestData<-empTest[c(MYinput, MYtarget)]


library(rpart)

myrpart<-rpart(STATUS~.,
               data=empTrain[,c(MYinput,MYtarget)],
               method="class",
               parms = list(split="information"),
               control=rpart.control(usesurrogate = 0, maxsurrogate = 0))
myrpart
print(myrpart)
printcp(myrpart)
fancyRpartPlot(myrpart, main="Decision Tree MFG10YearTerminationData $ STATUS")


#predict using decision train
mydt<-predict(myrpart, data=empTest[c(MYinput, MYtarget)], type="class")

#confusion matrix to check
#confusionMatrix(data=mydt, reference = empTest$STATUS,positive = "Yes", mode = "prec_recall")
#error


#table(empTest[c(MYinput,MYtarget)]$STATUS, mydt, dnn = c("Actual","Predicted"))

set.seed(crv$seed)

myrf<-randomForest(STATUS~.,
                   data=empTrain[,c(MYinput,MYtarget)],
                   ntree=500,
                   mtry=2,
                   importance=TRUE,
                   na.action = randomForest::na.roughfix,
                   replace=FALSE)
myrf



pROC::roc(myrf$y,as.numeric(myrf$predicted))


pROC::ci.auc(myrf$y,as.numeric(myrf$predicted))

rn<-round(randomForest::importance(myrf),2)
rn[order(rn[,3],decreasing = TRUE),]

varImpPlot(myrf, type = 1, main="Variable importance", sub="random forest model")