Case Study: HR Data (Analysis and Visualization)

Adarsh Adwait

---

Case Study

Case Study: HR Data (Analysis and Visualization)

Marketing Problem

Dataset

Consider a simple dataset of “Case Study: HR Data (Analysis and Visualization)”

Reading HR Emplyement Dataset

Dimensions of the Dataset

# printing Dimensions
dim(hrdata)

[1] 8995   17

Structure of the Dataset

# structure of the dataset
str(hrdata)

'data.frame':   8995 obs. of  17 variables:
 $ CandidateRef            : int  2110407 2112635 2112838 2115021 2115125 2117167 2119124 2127572 2138169 2143362 ...
 $ DOJExtended             : Factor w/ 2 levels "No","Yes": 2 1 1 1 2 2 2 2 1 1 ...
 $ DurationToAcceptOffer   : int  14 18 3 26 1 17 37 16 1 6 ...
 $ NoticePeriod            : int  30 30 45 30 120 30 30 0 30 30 ...
 $ OfferedBand             : Factor w/ 4 levels "E0","E1","E2",..: 3 3 3 3 3 2 3 2 2 2 ...
 $ PercentHikeExpectedInCTC: num  -20.8 50 42.8 42.8 42.6 ...
 $ PercentHikeOfferedInCTC : num  13.2 320 42.8 42.8 42.6 ...
 $ PercentDifferenceCTC    : num  42.9 180 0 0 0 ...
 $ JoiningBonus            : Factor w/ 2 levels "No","Yes": 1 1 1 1 1 1 1 1 1 1 ...
 $ CandidateRelocateActual : Factor w/ 2 levels "No","Yes": 1 1 1 1 2 1 1 1 1 1 ...
 $ Gender                  : Factor w/ 2 levels "Female","Male": 1 2 2 2 2 2 2 1 1 2 ...
 $ CandidateSource         : Factor w/ 3 levels "Agency","Direct",..: 1 3 1 3 3 3 3 2 3 3 ...
 $ RexInYrs                : int  7 8 4 4 6 2 7 8 3 3 ...
 $ LOB                     : Factor w/ 9 levels "AXON","BFSI",..: 5 8 8 8 8 8 8 7 2 3 ...
 $ Location                : Factor w/ 11 levels "Ahmedabad","Bangalore",..: 9 3 9 9 9 9 9 9 5 3 ...
 $ Age                     : int  34 34 27 34 34 34 32 34 26 34 ...
 $ Status                  : Factor w/ 2 levels "Joined","Not Joined": 1 1 1 1 1 1 1 1 1 1 ...

Descriptive Statistics of the Dataset

# structure of the dataset
psych::describe(hrdata)

                         vars    n       mean        sd  median    trimmed
CandidateRef                1 8995 2843647.38 486344.77 2807482 2831309.11
DOJExtended*                2 8995       1.47      0.50       1       1.46
DurationToAcceptOffer       3 8995      21.43     25.81      10      16.64
NoticePeriod                4 8995      39.29     22.22      30      37.91
OfferedBand*                5 8995       2.39      0.63       2       2.32
PercentHikeExpectedInCTC    6 8995      43.86     29.79      40      40.92
PercentHikeOfferedInCTC     7 8995      40.66     36.06      36      36.70
PercentDifferenceCTC        8 8995      -1.57     19.61       0      -3.08
JoiningBonus*               9 8995       1.05      0.21       1       1.00
CandidateRelocateActual*   10 8995       1.14      0.35       1       1.05
Gender*                    11 8995       1.83      0.38       2       1.91
CandidateSource*           12 8995       1.89      0.67       2       1.86
RexInYrs                   13 8995       4.24      2.55       4       4.01
LOB*                       14 8995       5.18      2.38       5       5.30
Location*                  15 8995       4.94      3.00       3       4.78
Age                        16 8995      29.91      4.10      29      29.86
Status*                    17 8995       1.19      0.39       1       1.11
                               mad        min        max      range  skew
CandidateRef             668050.66 2109586.00 3836076.00 1726490.00  0.16
DOJExtended*                  0.00       1.00       2.00       1.00  0.13
DurationToAcceptOffer        13.34       0.00     224.00     224.00  1.61
NoticePeriod                  0.00       0.00     120.00     120.00  0.74
OfferedBand*                  0.00       1.00       4.00       3.00  0.81
PercentHikeExpectedInCTC     19.53     -68.83     359.77     428.60  2.29
PercentHikeOfferedInCTC      20.76     -60.53     471.43     531.96  2.75
PercentDifferenceCTC          6.85     -67.27     300.00     367.27  4.80
JoiningBonus*                 0.00       1.00       2.00       1.00  4.31
CandidateRelocateActual*      0.00       1.00       2.00       1.00  2.03
Gender*                       0.00       1.00       2.00       1.00 -1.73
CandidateSource*              0.00       1.00       3.00       2.00  0.13
RexInYrs                      1.48       0.00      24.00      24.00  1.29
LOB*                          4.45       1.00       9.00       8.00 -0.22
Location*                     1.48       1.00      11.00      10.00  0.50
Age                           4.45      20.00      60.00      40.00  0.42
Status*                       0.00       1.00       2.00       1.00  1.61
                         kurtosis      se
CandidateRef                -1.46 5127.95
DOJExtended*                -1.98    0.01
DurationToAcceptOffer        2.65    0.27
NoticePeriod                 0.79    0.23
OfferedBand*                 0.30    0.01
PercentHikeExpectedInCTC    12.44    0.31
PercentHikeOfferedInCTC     15.92    0.38
PercentDifferenceCTC        45.55    0.21
JoiningBonus*               16.61    0.00
CandidateRelocateActual*     2.14    0.00
Gender*                      1.01    0.00
CandidateSource*            -0.81    0.01
RexInYrs                     3.20    0.03
LOB*                        -1.23    0.03
Location*                   -1.56    0.03
Age                          0.34    0.04
Status*                      0.58    0.00

```

DISCRETE DATA DISTRIBUTION

(ONE-WAY, TWO-WAY AND THREE-WAY CONTINGENCY TABLES)

Percentage of the candidates (Joined / Not joined)

round(prop.table(with(dataset, table(Status)))*100,2)

Status
    Joined Not Joined 
      81.3       18.7

Percentage of the candidates (Joined / Not joined) by DOJ Extended

mytable <- xtabs(~ DOJExtended+Status, data=dataset)
addmargins(round(prop.table(mytable,1)*100,2),1)

           Status
DOJExtended Joined Not Joined
        No   81.08      18.92
        Yes  81.55      18.45
        Sum 162.63      37.37

Number of the candidates (Joined / Not joined) by Notice Period

mytable <- xtabs(~ NoticePeriod+Status, data=dataset)
addmargins(mytable)

            Status
NoticePeriod Joined Not Joined  Sum
         0      726         51  777
         30    4393        765 5158
         45     397        129  526
         60    1285        470 1755
         75      75         35  110
         90     415        212  627
         120     22         20   42
         Sum   7313       1682 8995

Percentage of the candidates (Joined / Not joined) by Notice Period

mytable <- xtabs(~ NoticePeriod+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

            Status
NoticePeriod Joined Not Joined
         0    93.44       6.56
         30   85.17      14.83
         45   75.48      24.52
         60   73.22      26.78
         75   68.18      31.82
         90   66.19      33.81
         120  52.38      47.62
         Sum 514.05     185.95

Percentage of the candidates (Joined / Not joined) by Joining Bonus

mytable <- xtabs(~ JoiningBonus+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

            Status
JoiningBonus Joined Not Joined
         No   81.34      18.66
         Yes  80.58      19.42
         Sum 161.91      38.09

Percentage of the candidates (Joined / Not joined) by Gender

mytable <- xtabs(~ Gender+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

        Status
Gender   Joined Not Joined
  Female  82.40      17.60
  Male    81.07      18.93
  Sum    163.47      36.53

Percentage of the candidates (Joined / Not joined) by Candidate Source

mytable <- xtabs(~ CandidateSource+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

                   Status
CandidateSource     Joined Not Joined
  Agency             75.82      24.18
  Direct             82.00      18.00
  Employee Referral  88.00      12.00
  Sum               245.83      54.17

Percentage of the candidates (Joined / Not joined) by Offered Band

mytable <- xtabs(~ OfferedBand+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

           Status
OfferedBand Joined Not Joined
        E0   76.30      23.70
        E1   81.30      18.70
        E2   80.97      19.03
        E3   85.15      14.85
        Sum 323.72      76.28

Percentage of the candidates (Joined / Not joined) by Line of Business (LOB)

mytable <- xtabs(~ LOB+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

            Status
LOB          Joined Not Joined
  AXON        77.46      22.54
  BFSI        75.86      24.14
  CSMP        81.52      18.48
  EAS         73.41      26.59
  ERS         78.11      21.89
  ETS         83.07      16.93
  Healthcare  82.26      17.74
  INFRA       87.79      12.21
  MMS        100.00       0.00
  Sum        739.48     160.52

Number of the candidates (Joined / Not joined) by Location

mytable <- xtabs(~ Location+Status, data=dataset)
addmargins(mytable)

           Status
Location    Joined Not Joined  Sum
  Ahmedabad      5          1    6
  Bangalore   1742        488 2230
  Chennai     2486        664 3150
  Cochin         7          1    8
  Gurgaon      118         28  146
  Hyderabad    266         75  341
  Kolkata      100         29  129
  Mumbai       176         21  197
  Noida       2362        365 2727
  Others        13          0   13
  Pune          38         10   48
  Sum         7313       1682 8995

Percentage of the candidates (Joined / Not joined) by Location

mytable <- xtabs(~ Location+Status, data=dataset)
round(addmargins(prop.table(mytable,1)*100,1),2)

           Status
Location    Joined Not Joined
  Ahmedabad  83.33      16.67
  Bangalore  78.12      21.88
  Chennai    78.92      21.08
  Cochin     87.50      12.50
  Gurgaon    80.82      19.18
  Hyderabad  78.01      21.99
  Kolkata    77.52      22.48
  Mumbai     89.34      10.66
  Noida      86.62      13.38
  Others    100.00       0.00
  Pune       79.17      20.83
  Sum       919.34     180.66

CONTINUOUS DATA DISTRIBUTION

(SUMMARY TABLES)

Average Age of the Candidates (Joined / Not joined)

result<-aggregate(dataset$Age, by=list(Status=dataset$Status), FUN=mean)
names(result)[2]<-"AverageAgeofCandidates"
result$AverageAgeofCandidates<-round(result$AverageAgeofCandidates, 2)
result

      Status AverageAgeofCandidates
1     Joined                  30.00
2 Not Joined                  29.52

Average Age and Notice Period of the Candidates (Joined / Not joined)

result<-aggregate(dataset[c("Age","NoticePeriod")], by=list(Status=dataset$Status), FUN=mean)
names(result)[2:3]<-c("AverageAgeofCandidates","AverageNoticePeriod")
result[2:3]<-round(result[2:3],2)
result

      Status AverageAgeofCandidates AverageNoticePeriod
1     Joined                  30.00               37.24
2 Not Joined                  29.52               48.19

Average Age of the Candidates (Joined / Not joined) by Gender (Male / Female)

result<-aggregate(dataset$Age, by=list(Status=dataset$Status,Gender=dataset$Gender), FUN=mean)
names(result)[3]<-"AverageAgeofCandidates"
result$AverageAgeofCandidates<-round(result$AverageAgeofCandidates, 2)
result

      Status Gender AverageAgeofCandidates
1     Joined Female                  29.09
2 Not Joined Female                  28.02
3     Joined   Male                  30.20
4 Not Joined   Male                  29.81

Average (Age and Relevant Years of Experience) of the candidates (Joined / Not joined) by Gender (Male / Female)

result<-aggregate(dataset[c("Age","RexInYrs")], by=list(Status=dataset$Status,Gender=dataset$Gender), FUN=mean)
names(result)[3:4]<-c("AverageAgeofCandidates","YearsOfExperience")
result[3:4]<-round(result[3:4],2)
result

      Status Gender AverageAgeofCandidates YearsOfExperience
1     Joined Female                  29.09              3.45
2 Not Joined Female                  28.02              3.61
3     Joined   Male                  30.20              4.35
4 Not Joined   Male                  29.81              4.60

Average (Age, Relevant Years of Experience and Number of days taken by the candidate to accept the offer) of candidates (Joined / Not joined) by Gender (Male / Female)

result<-aggregate(dataset[c("Age","RexInYrs","DurationToAcceptOffer")], by=list(Status=dataset$Status,Gender=dataset$Gender), FUN=mean)
names(result)[3:4]<-c("AverageAgeofCandidates","YearsOfExperience")
result[3:5]<-round(result[3:5],2)
result

      Status Gender AverageAgeofCandidates YearsOfExperience
1     Joined Female                  29.09              3.45
2 Not Joined Female                  28.02              3.61
3     Joined   Male                  30.20              4.35
4 Not Joined   Male                  29.81              4.60
  DurationToAcceptOffer
1                 19.43
2                 24.10
3                 20.88
4                 25.12

Average (Age and Notice Period) of the candidates (Joined / Not joined) by Gender (Male / Female)

result<-aggregate(dataset[c("Age","NoticePeriod")], by=list(Status=dataset$Status,Gender=dataset$Gender), FUN=mean)
names(result)[3]<-"AverageAgeofCandidates"
result[3:4]<-round(result[3:4],2)
result

      Status Gender AverageAgeofCandidates NoticePeriod
1     Joined Female                  29.09        35.48
2 Not Joined Female                  28.02        46.65
3     Joined   Male                  30.20        37.62
4 Not Joined   Male                  29.81        48.49

Visualization of the Dataset

# Pie Chart from data frame with Appended Sample Sizes
mytable <- table(Status)
pie(mytable, 
   main="Pie Chart of Percentage of Candidates who Joined / Not Joined")

plot of chunk unnamed-chunk-22

Visualization of Joining Status vs Gender

            Gender
Status       Female Male
  Joined       1278 6035
  Not Joined    273 1409

plot of chunk unnamed-chunk-23

Visualization of Joining Status vs Offered Brand

            OfferedBand
Status         E0   E1   E2   E3
  Joined      161 4527 2195  430
  Not Joined   50 1041  516   75

plot of chunk unnamed-chunk-24

Visualization of Joining Status vs Joining Bonus

            JoiningBonus
Status         No  Yes
  Joined     6977  336
  Not Joined 1601   81

plot of chunk unnamed-chunk-25

Visualization of Joining Status vs Referral Source

            CandidateSource
Status       Agency Direct Employee Referral
  Joined       1960   3937              1416
  Not Joined    625    864               193

plot of chunk unnamed-chunk-26

Visualization of Joining Status vs City Location

            Location
Status       Ahmedabad Bangalore Chennai Cochin Gurgaon Hyderabad Kolkata
  Joined             5      1742    2486      7     118       266     100
  Not Joined         1       488     664      1      28        75      29
            Location
Status       Mumbai Noida Others Pune
  Joined        176  2362     13   38
  Not Joined     21   365      0   10

plot of chunk unnamed-chunk-27

Visualization of Joining Status vs Line of Business

            LOB
Status       AXON BFSI CSMP  EAS  ERS  ETS Healthcare INFRA  MMS
  Joined      440 1059  472  254 1895  574        102  2502   15
  Not Joined  128  337  107   92  531  117         22   348    0

plot of chunk unnamed-chunk-28

Boxplot of Age by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-29

Boxplot of Duration to Accept Offer by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-30

Boxplot of Notice Period by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-31

Boxplot of Percent Hike (CTC) Expected by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-32

Boxplot of Percent Hike (CTC) Offered by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-33

Boxplot of Experience by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-34

Scatter Plot of Experience and Duration to Accept Offer by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-35

Scatter Plot of Experience and Notice Period by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-36

Scatter Plot of Experience and Percent Hike (CTC) Expected by Candidate by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-37

Scatter Plot of Experience and Percent Hike (CTC) Offered by Candidate by Status (Joined / Not Joined)

plot of chunk unnamed-chunk-38

Split your data into test and train

library(caTools)
library(caret)
# use set.seed to use the same random number sequence
set.seed(123)
# craeting 75% data for training 
split <- sample.split(Status, SplitRatio = 0.75)
trainData <- subset(hrdata, split == TRUE)
# dimensions of training data
dim(trainData)

[1] 6747   17

# creating 25% data for testing
testData <- subset(hrdata, split == FALSE)
# dimensions of testing data
dim(testData)

[1] 2248   17

CLASSIFICATION USING BINOMIAL LOGISTIC MODEL

(Classification using Logistic Regression with glm())

Classifier 1


Call:
glm(formula = Status ~ PercentDifferenceCTC + DOJExtended + OfferedBand + 
    NoticePeriod + RexInYrs + LOB + CandidateRelocateActual, 
    family = binomial(), data = trainData)

Deviance Residuals: 
     Min        1Q    Median        3Q       Max  
-1.50722  -0.69675  -0.52765  -0.00012   2.50626  

Coefficients:
                             Estimate Std. Error z value Pr(>|z|)    
(Intercept)                -5.281e-01  2.512e-01  -2.102 0.035526 *  
PercentDifferenceCTC       -2.731e-03  1.975e-03  -1.383 0.166753    
DOJExtendedYes             -2.085e-01  6.970e-02  -2.992 0.002773 ** 
OfferedBandE1              -1.288e+00  2.121e-01  -6.075 1.24e-09 ***
OfferedBandE2              -1.257e+00  2.330e-01  -5.397 6.79e-08 ***
OfferedBandE3              -1.662e+00  3.082e-01  -5.392 6.98e-08 ***
NoticePeriod                2.169e-02  1.525e-03  14.229  < 2e-16 ***
RexInYrs                    2.069e-02  2.057e-02   1.006 0.314503    
LOBBFSI                    -3.553e-01  1.515e-01  -2.346 0.018974 *  
LOBCSMP                    -4.261e-01  1.842e-01  -2.314 0.020685 *  
LOBEAS                      2.296e-01  1.982e-01   1.159 0.246606    
LOBERS                     -9.486e-02  1.434e-01  -0.662 0.508159    
LOBETS                     -5.960e-01  1.801e-01  -3.310 0.000934 ***
LOBHealthcare              -4.420e-01  3.074e-01  -1.438 0.150429    
LOBINFRA                   -8.630e-01  1.572e-01  -5.490 4.03e-08 ***
LOBMMS                     -1.771e+01  2.131e+03  -0.008 0.993368    
CandidateRelocateActualYes -1.729e+01  2.008e+02  -0.086 0.931360    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 6502.9  on 6746  degrees of freedom
Residual deviance: 5693.1  on 6730  degrees of freedom
AIC: 5727.1

Number of Fisher Scoring iterations: 17

Subset of statistically significant variables

  1. Offered Band
  2. Notice Period
  3. LOB
  4. DOJ Extended

Prediction using classifier 1

# prediction usinf classifier 1
predProbClass1 <- predict(logitClassifier1, type = 'response', newdata = testData[-17])
yPred1 <- ifelse(predProbClass1 > 0.5, "Joined", "Not Joined")
table(yPred1)

yPred1
    Joined Not Joined 
        54       2194

AIC of the classifiers

AIC(logitClassifier1)

[1] 5727.065

Confusion matrix using classifier 1

library(MLmetrics)
ConfusionMatrix(yPred1, testData[, 17])

            y_pred
y_true       Joined Not Joined
  Joined         31       1797
  Not Joined     23        397

Accuracy using classifier 1

Accuracy(yPred1, testData$Status)

[1] 0.1903915

Sensitivity using classifier 1

Sensitivity(yPred1, testData$Status, positive = "Joined")

[1] 0.5740741

Specificity using classifier 1

Specificity(yPred1, testData$Status, positive = "Joined")

[1] 0.180948

ROC Plot using Classifier 1

plot of chunk unnamed-chunk-47