R Data Science

library(openxlsx)
HR_Analytics <- read.xlsx("H:/Yashwanth/Kaggle/HR_comma_sep.xlsx",sheet = 1)
# lsf.str("package:openxlsx"); help(package="openxlsx")

Gradient Boosting

Objective : H(x) = {Set of all possible hypothesis} Reference

# Summary
rbind(Workaccident=prop.table(table(HR_Analytics$Work_accident))*100,
      Promotion=prop.table(table(HR_Analytics$promotion_last_5years))*100,
      Left=prop.table(table(HR_Analytics$left))*100)

                    0         1
Workaccident 85.53904 14.460964
Promotion    97.87319  2.126808
Left         76.19175 23.808254

Basic summary(Insights) of the data:

1. ~14% of employees falls into Workaccident bucket
2. ~2% of employees falls into promotion bucket
3. ~23% of employees falls into left bucket

Distribution of Variables

# Distribution
View(head(HR_Analytics))
library(ggplot2)
attach(HR_Analytics)
ggplot(HR_Analytics,aes(x=satisfaction_level)) +
  facet_wrap(~ left)+
  geom_histogram(aes(y=..density..),col = "blue2") +
  stat_function(fun = dnorm,args = list(mean=mean(satisfaction_level),sd=sd(satisfaction_level)),colour = "red") +
  labs(title = "satisfaction_level Distribution")

ggplot(HR_Analytics,aes(x=last_evaluation)) +
  facet_wrap(~ left)+
  geom_histogram(aes(y=..density..),col = "blue2") +
  stat_function(fun = dnorm,args = list(mean=mean(last_evaluation),sd=sd(last_evaluation)),colour = "red") +
  labs(title = "last_evaluation Distribution")

ggplot(HR_Analytics,aes(x=number_project)) +
  facet_wrap(~ left)+
  geom_histogram(aes(y=..density..),col = "blue2") +
  stat_function(fun = dnorm,args = list(mean=mean(number_project),sd=sd(number_project)),colour = "red") +
  labs(title = "number_project Distribution")

ggplot(HR_Analytics,aes(x=average_montly_hours)) +
  facet_wrap(~ left)+
  geom_histogram(aes(y=..density..),col = "blue2") +
  stat_function(fun = dnorm,args = list(mean=mean(average_montly_hours),sd=sd(average_montly_hours)),
                colour = "red") +
  labs(title = "average_montly_hours Distribution")

ggplot(HR_Analytics,aes(x=time_spend_company)) +
  facet_wrap(~ left)+
  geom_histogram(aes(y=..density..),col = "blue2") +
  stat_function(fun = dnorm,args = list(mean=mean(time_spend_company),sd=sd(time_spend_company)),colour = "red") +
  labs(title = "time_spend_company Distribution")

detach(HR_Analytics)