data <- data %>%
  mutate_if(is.character, as.factor) %>%
  select(Attrition, everything())

# Attrition berdasarkan Income Level
plot1 <- ggplot(data, aes(x = MonthlyIncome, fill = Attrition)) + 
  geom_density(alpha = 0.6) +
  labs(x = "Monthly Income", y = "") +
  ggtitle("Attrition by income level") +
  theme_classic()

#Attrition berdasarkan Department
# let's see what percentage of people are leaving from their departments
dept <- data %>% 
        group_by(Department, Attrition) %>% 
        summarise(n = n())

# creating groups and summing by group
dept$totals <- ifelse(dept$Department == "Human Resources", 
                      sum(data$Department == "Human Resources"), 
                      ifelse(dept$Department == "Research & Development", 
                             sum(data$Department == "Research & Development"),
                             ifelse(dept$Department == "Sales",
                                    sum(data$Department == "Sales"), data$Department))) 

# using dplyr's mutate() function to create the variable 'attr_perc'
dept <- dept %>% 
        mutate(attr_perc = n / totals * 100)

plot2 <- dept %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = reorder(Department, attr_perc), y = attr_perc, fill = Department)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="", y="% of employees")+
      ggtitle("Attrition % by Department") +
      theme_classic()

# Attrition by Amount of Business Travel

# do people who travel more for business leave at higher rates?
trav_attr <- data %>% 
                group_by(BusinessTravel, Attrition) %>% 
                summarise(n = n())

# creating traveler groups
trav_attr$totals <- ifelse(trav_attr$BusinessTravel == "Non-Travel", 
                           sum(data$BusinessTravel == "Non-Travel"), 
                           ifelse(trav_attr$BusinessTravel == "Travel_Frequently", 
                                  sum(data$BusinessTravel == "Travel_Frequently"),
                                  ifelse(trav_attr$BusinessTravel == "Travel_Rarely",
                                         sum(data$BusinessTravel == "Travel_Rarely"), trav_attr$BusinessTravel))) 

trav_attr <- trav_attr %>% 
                mutate(attr_perc = n / totals * 100)

plot3 <- trav_attr %>% 
        filter(Attrition == "Yes") %>% 
          ggplot(aes(x = reorder(BusinessTravel, attr_perc), y = attr_perc, fill = BusinessTravel)) + 
          geom_bar(stat = "identity", alpha = 0.9) +
          labs(x="", y="% of employees") +
          ggtitle("Attrition % by Amount of Business Travel") +
          theme_classic()

# attrition by education level
edu_attr <- data %>% 
            group_by(Education, Attrition) %>% 
            summarise(n = n())

edu_tots <- data %>% 
            group_by(Education) %>% 
            summarise(num = n())

edu_attr <- left_join(edu_attr, edu_tots, by = "Education")

edu_attr <- edu_attr %>% mutate(attr_perc = n / num * 100)

plot4 <- edu_attr %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = reorder(Education, attr_perc), y = attr_perc, fill = Education)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="Education Level", y="% of employees") +
      ggtitle("Attrition % by Education Level") +
      theme_classic()

# Attrition by MaritalStatus

mar_attr <- data %>% 
            group_by(MaritalStatus, Attrition) %>% 
            summarise(n = n())

mar_tots <- data %>% 
            group_by(MaritalStatus) %>% 
            summarise(nums = n())

mar_attr <- left_join(mar_attr, mar_tots, by = "MaritalStatus")

mar_attr <- mar_attr %>% 
            mutate(attr_perc = n / nums * 100)

plot5 <- mar_attr %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = reorder(MaritalStatus, attr_perc), y = attr_perc, fill = MaritalStatus)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="", y="% of employees") +
      ggtitle("Attrition % by Marital Status") +
      theme_classic()

# attrition by gender and marital status
gend_attr <- data %>% 
            group_by(Gender, MaritalStatus, Attrition) %>% 
            summarise(n = n())

gend_tots <- data %>% 
            group_by(Gender, MaritalStatus) %>% 
            summarise(nums = n())

gend_attr <- left_join(gend_attr, gend_tots)

gend_attr <- gend_attr %>% 
                mutate(attr_perc = n / nums * 100)

gend_attr$type <- paste(gend_attr$MaritalStatus, gend_attr$Gender, sep = " ")

plot6 <- gend_attr %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = reorder(type, attr_perc), y = attr_perc, fill = Gender)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="", y="% of employees") +
      ggtitle("Attrition % by Gender and Marital Status") +
      theme_classic() +
      coord_flip()

# attrition by JobInvolvement

jobinv <- data %>% 
        group_by(JobInvolvement, Attrition) %>% 
        summarise(n = n())

jobtots <- data %>% 
        group_by(JobInvolvement) %>% 
        summarise(nums = n())

jobinv <- left_join(jobinv, jobtots, by = "JobInvolvement")

jobinv <- jobinv %>% 
            mutate(attr_perc = n / nums * 100)

plot7 <- jobinv %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = JobInvolvement, y = attr_perc, fill = JobInvolvement)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="Job Involvement Level", y="% of employee attrition") +
      ggtitle("Attrition % by Job Involvement") +
      theme_classic()

# attrition by overtime

ovt <- data %>% 
        group_by(OverTime, Attrition) %>% 
        summarise(n = n())

ovttots <- data %>% 
            group_by(OverTime) %>% 
            summarise(nums = n())

ovt <- left_join(ovt, ovttots, by = "OverTime")

ovt <- ovt %>% 
        mutate(attr_perc = n / nums * 100)

plot8 <- ovt %>% 
    filter(Attrition == "Yes") %>% 
      ggplot(aes(x = OverTime, y = attr_perc, fill = OverTime)) + 
      geom_bar(stat = "identity", alpha = 0.9) +
      labs(x="Overtime", y="% of employee attrition") +
      ggtitle("Attrition % by Overtime (Y/N)") +
      theme_classic()

ggarrange(plot1, plot2, plot3, plot4, plot5, plot6, plot7, plot8, ncol = 2, nrow = 2)

# attrition by years worked for company

yrcomp <- data %>% 
            group_by(YearsAtCompany) %>% 
            filter(Attrition == "Yes") %>% 
            summarise(n = n())

yrcomp$Attrition <- "Left"

yrno <- data %>% 
        group_by(YearsAtCompany) %>% 
        filter(Attrition == "No") %>% 
        summarise(n = n())

yrno$Attrition <- "Stayed"

yearscomp <- rbind(yrcomp, yrno)

plot9 <- ggplot(yearscomp, aes(x = YearsAtCompany, y = n, fill = Attrition)) +
  geom_bar(position = "stack", stat = "identity", alpha = 0.9) +
  xlab("Years With Company") +
  ylab("# of employees") +
  ggtitle("Attrition by years worked for company") +
  theme_classic()

# attrition by number of years in current role
yrscomp <- data %>% 
            group_by(YearsInCurrentRole) %>% 
            filter(Attrition == "Yes") %>% 
            summarise(n = n())

yrscomp$Attrition <- "Left"

yrsno <- data %>% 
            group_by(YearsInCurrentRole) %>% 
            filter(Attrition == "No") %>% 
            summarise(n = n())

yrsno$Attrition <- "Stayed"

jyearscomp <- rbind(yrscomp, yrsno)

plot10 <- ggplot(jyearscomp, aes(x = YearsInCurrentRole, y = n, fill = Attrition)) +
  geom_bar(position = "stack", stat = "identity", alpha = 0.9) +
  xlab("Years In Current Role") +
  ylab("# of employees") +
  ggtitle("Attrition by years in current role") +
  theme_classic()

# attrition by number of years with current manager

mancomp <- data %>% 
            group_by(YearsWithCurrManager) %>% 
            filter(Attrition == "Yes") %>% 
            summarise(n = n())

mancomp$Attrition <- "Left"

manno <- data %>% 
            group_by(YearsWithCurrManager) %>% 
            filter(Attrition == "No") %>% 
            summarise(n = n())

manno$Attrition <- "Stayed"

managercomp <- rbind(mancomp, manno)

plot11 <- ggplot(managercomp, aes(x = YearsWithCurrManager, y = n, fill = Attrition)) +
  geom_bar(position = "stack", stat = "identity", alpha = 0.9) +
  xlab("Years With Current Manager") +
  ylab("# of employees") +
  ggtitle("Attrition by years with current manager") +
  theme_classic()

ggarrange(plot9, plot10, plot11, ncol = 2, nrow= 2)

sm1 <- surv_summary(model_comp, data=data)

datatable(sm1,
          caption = htmltools::tags$caption(
            style = 'caption-side: bottom; text-align: center,',
            htmltools::em('Summary of Survival Curves.')))

summary(model_comp)$table

ggsurvplot(model_comp,
           data = data,
           pval = TRUE,
           conf.int = TRUE,
           xlab = "Time in Years",
           ggtheme = theme_bw(),
           risk.table = "abs_pct",
           risk.table.y.text.col = TRUE,
           risk.table.y.text = FALSE,
           ncensor.plot = TRUE,
           surv.median.line = "hv",
           legend.labs = c("Overtime", "No Overtime"),
           palette = c("#2E9FDF", "#9B59B6"))

surv_diff <- survdiff(Surv(time, event) ~ group, data=data)
surv_diff