[go: up one dir, main page]
Scribd
Upload
53 views5 pages

Assignment 2

The document describes analyzing employee attrition data using different regression models. It loads attrition data, cleans and structures the data, then fits linear regression, logistic regression, and probit regression models to predict attrition. Key predictors included age, job satisfaction, income, years with company, and since last promotion. The logistic regression models had slightly better fit than the linear models based on residual deviance and AIC values.

Uploaded by

Niharika Pradhan I H21O32
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
53 views5 pages

Assignment 2

The document describes analyzing employee attrition data using different regression models. It loads attrition data, cleans and structures the data, then fits linear regression, logistic regression, and probit regression models to predict attrition. Key predictors included age, job satisfaction, income, years with company, and since last promotion. The logistic regression models had slightly better fit than the linear models based on residual deviance and AIC values.

Uploaded by

Niharika Pradhan I H21O32
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 5

> library(readxl)

> For.R = read.csv("G:\\My Drive\\XLRI\\Term 4\\Applied Econometrics\\Assignment


2\\Assignment 2_Data.csv")
> str(For.R)
'data.frame': 1470 obs. of 13 variables:
$ Age : int 41 49 37 33 27 32 59 30 38 36 ...
$ Gender : chr "Female" "Male" "Male" "Female" ...
$ JobSatisfaction : int 4 2 3 3 2 4 1 3 3 3 ...
$ MonthlyIncome : int 5993 5130 2090 2909 3468 3068 2670 2693 9526
5237 ...
$ PercentSalaryHike : int 11 23 15 11 12 13 20 22 21 13 ...
$ PerformanceRating : int 3 4 3 3 3 3 4 4 4 3 ...
$ TotalWorkingYears : int 8 10 7 8 6 8 12 1 10 17 ...
$ TrainingTimesLastYear : int 0 3 3 3 3 2 3 2 2 3 ...
$ YearsAtCompany : int 6 10 0 8 2 7 1 1 9 7 ...
$ YearsInCurrentRole : int 4 7 0 7 2 7 0 0 7 7 ...
$ YearsSinceLastPromotion: int 0 1 0 3 2 3 0 0 1 7 ...
$ YearsWithCurrManager : int 5 7 0 0 2 6 0 0 8 7 ...
$ Attrition : chr "Yes" "No" "Yes" "No" ...
> For.R$Gender<-ifelse(For.R$Gender=="Female",1,0)
> For.R$Attrition<-ifelse(For.R$Attrition=="No",1,0)
> str(For.R)
'data.frame': 1470 obs. of 13 variables:
$ Age : int 41 49 37 33 27 32 59 30 38 36 ...
$ Gender : num 1 0 0 1 0 0 1 0 0 0 ...
$ JobSatisfaction : int 4 2 3 3 2 4 1 3 3 3 ...
$ MonthlyIncome : int 5993 5130 2090 2909 3468 3068 2670 2693 9526
5237 ...
$ PercentSalaryHike : int 11 23 15 11 12 13 20 22 21 13 ...
$ PerformanceRating : int 3 4 3 3 3 3 4 4 4 3 ...
$ TotalWorkingYears : int 8 10 7 8 6 8 12 1 10 17 ...
$ TrainingTimesLastYear : int 0 3 3 3 3 2 3 2 2 3 ...
$ YearsAtCompany : int 6 10 0 8 2 7 1 1 9 7 ...
$ YearsInCurrentRole : int 4 7 0 7 2 7 0 0 7 7 ...
$ YearsSinceLastPromotion: int 0 1 0 3 2 3 0 0 1 7 ...
$ YearsWithCurrManager : int 5 7 0 0 2 6 0 0 8 7 ...
$ Attrition : num 0 1 0 1 1 1 1 1 1 1 ...
> model1 = lm(For.R$Attrition ~ For.R$Age + For.R$Gender + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$PercentSalaryHike + For.R$PerformanceRating +
For.R$TotalWorkingYears + For.R$TrainingTimesLastYear + For.R$YearsAtCompany +
For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager)
> summary(model1)

Call:
lm(formula = For.R$Attrition ~ For.R$Age + For.R$Gender + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$PercentSalaryHike + For.R$PerformanceRating +
For.R$TotalWorkingYears + For.R$TrainingTimesLastYear + For.R$YearsAtCompany +
For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager)

Residuals:
Min 1Q Median 3Q Max
-1.09700 0.01972 0.13110 0.20845 0.38821

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 4.892e-01 1.055e-01 4.635 3.90e-06 ***
For.R$Age 4.091e-03 1.419e-03 2.883 0.003991 **
For.R$Gender 1.496e-02 1.897e-02 0.789 0.430416
For.R$JobSatisfaction 3.560e-02 8.424e-03 4.226 2.53e-05 ***
For.R$MonthlyIncome 6.789e-06 3.128e-06 2.170 0.030135 *
For.R$PercentSalaryHike 3.872e-03 4.010e-03 0.966 0.334382
For.R$PerformanceRating -3.697e-02 4.068e-02 -0.909 0.363690
For.R$TotalWorkingYears 7.120e-04 2.509e-03 0.284 0.776632
For.R$TrainingTimesLastYear 1.846e-02 7.206e-03 2.562 0.010511 *
For.R$YearsAtCompany -4.747e-03 3.146e-03 -1.509 0.131610
For.R$YearsInCurrentRole 1.341e-02 4.185e-03 3.205 0.001380 **
For.R$YearsSinceLastPromotion -1.247e-02 3.713e-03 -3.359 0.000801 ***
For.R$YearsWithCurrManager 1.288e-02 4.306e-03 2.991 0.002824 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3552 on 1457 degrees of freedom


Multiple R-squared: 0.07523, Adjusted R-squared: 0.06762
F-statistic: 9.878 on 12 and 1457 DF, p-value: < 2.2e-16

> model2 = lm(For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +


For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager)
> summary(model2)

Call:
lm(formula = For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager)

Residuals:
Min 1Q Median 3Q Max
-1.13679 0.02238 0.13368 0.20655 0.38999

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 4.369e-01 5.049e-02 8.652 < 2e-16 ***
For.R$Age 4.254e-03 1.171e-03 3.632 0.000291 ***
For.R$JobSatisfaction 3.522e-02 8.407e-03 4.189 2.97e-05 ***
For.R$MonthlyIncome 6.207e-06 2.417e-06 2.568 0.010318 *
For.R$TrainingTimesLastYear 1.845e-02 7.188e-03 2.566 0.010379 *
For.R$YearsInCurrentRole 1.108e-02 3.868e-03 2.865 0.004233 **
For.R$YearsSinceLastPromotion -1.416e-02 3.562e-03 -3.974 7.42e-05 ***
For.R$YearsWithCurrManager 9.897e-03 3.809e-03 2.599 0.009454 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3551 on 1462 degrees of freedom


Multiple R-squared: 0.07271, Adjusted R-squared: 0.06827
F-statistic: 16.38 on 7 and 1462 DF, p-value: < 2.2e-16

> logitmodel1 = glm(For.R$Attrition ~ For.R$Age + For.R$Gender +


For.R$JobSatisfaction + For.R$MonthlyIncome + For.R$PercentSalaryHike +
For.R$PerformanceRating + For.R$TotalWorkingYears + For.R$TrainingTimesLastYear +
For.R$YearsAtCompany + For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager, family = 'binomial')
> summary(logitmodel1)

Call:
glm(formula = For.R$Attrition ~ For.R$Age + For.R$Gender + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$PercentSalaryHike + For.R$PerformanceRating +
For.R$TotalWorkingYears + For.R$TrainingTimesLastYear + For.R$YearsAtCompany +
For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager, family = "binomial")

Deviance Residuals:
Min 1Q Median 3Q Max
-3.0095 0.2713 0.4663 0.6515 1.2068

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.044e+00 8.377e-01 -1.247 0.212517
For.R$Age 3.027e-02 1.155e-02 2.621 0.008761 **
For.R$Gender 1.212e-01 1.539e-01 0.788 0.430974
For.R$JobSatisfaction 2.794e-01 6.701e-02 4.170 3.04e-05 ***
For.R$MonthlyIncome 8.042e-05 3.048e-05 2.639 0.008325 **
For.R$PercentSalaryHike 3.617e-02 3.256e-02 1.111 0.266669
For.R$PerformanceRating -3.265e-01 3.285e-01 -0.994 0.320229
For.R$TotalWorkingYears 7.739e-03 2.249e-02 0.344 0.730711
For.R$TrainingTimesLastYear 1.515e-01 5.943e-02 2.549 0.010815 *
For.R$YearsAtCompany -3.670e-02 3.111e-02 -1.180 0.238110
For.R$YearsInCurrentRole 1.305e-01 3.865e-02 3.376 0.000736 ***
For.R$YearsSinceLastPromotion -1.413e-01 3.573e-02 -3.954 7.70e-05 ***
For.R$YearsWithCurrManager 1.195e-01 3.874e-02 3.085 0.002034 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 1298.6 on 1469 degrees of freedom


Residual deviance: 1172.2 on 1457 degrees of freedom
AIC: 1198.2

Number of Fisher Scoring iterations: 5

> logitmodel2 = lm(For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +


For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager, family = 'binomial')
Warning message:
In lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) :
extra argument ‘family’ will be disregarded
> logitmodel2 = glm(For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager, family = 'binomial')
> summary(logitmodel2)

Call:
glm(formula = For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager,
family = "binomial")

Deviance Residuals:
Min 1Q Median 3Q Max
-3.1051 0.2772 0.4720 0.6521 1.2166

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.502e+00 4.003e-01 -3.752 0.000175 ***
For.R$Age 3.200e-02 9.690e-03 3.303 0.000957 ***
For.R$JobSatisfaction 2.751e-01 6.669e-02 4.125 3.71e-05 ***
For.R$MonthlyIncome 7.757e-05 2.456e-05 3.158 0.001588 **
For.R$TrainingTimesLastYear 1.526e-01 5.905e-02 2.584 0.009769 **
For.R$YearsInCurrentRole 1.140e-01 3.643e-02 3.130 0.001748 **
For.R$YearsSinceLastPromotion -1.574e-01 3.337e-02 -4.718 2.38e-06 ***
For.R$YearsWithCurrManager 1.001e-01 3.540e-02 2.828 0.004690 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 1298.6 on 1469 degrees of freedom


Residual deviance: 1175.5 on 1462 degrees of freedom
AIC: 1191.5

Number of Fisher Scoring iterations: 5

> probitmodel1 = glm(For.R$Attrition ~ For.R$Age + For.R$Gender +


For.R$JobSatisfaction + For.R$MonthlyIncome + For.R$PercentSalaryHike +
For.R$PerformanceRating + For.R$TotalWorkingYears + For.R$TrainingTimesLastYear +
For.R$YearsAtCompany + For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager, family = binomial(link = "probit"))
> summary(probitmodel1)

Call:
glm(formula = For.R$Attrition ~ For.R$Age + For.R$Gender + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$PercentSalaryHike + For.R$PerformanceRating +
For.R$TotalWorkingYears + For.R$TrainingTimesLastYear + For.R$YearsAtCompany +
For.R$YearsInCurrentRole + For.R$YearsSinceLastPromotion +
For.R$YearsWithCurrManager, family = binomial(link = "probit"))

Deviance Residuals:
Min 1Q Median 3Q Max
-3.0559 0.2625 0.4802 0.6611 1.1496

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -4.776e-01 4.676e-01 -1.021 0.307077
For.R$Age 1.498e-02 6.338e-03 2.364 0.018083 *
For.R$Gender 6.800e-02 8.515e-02 0.799 0.424517
For.R$JobSatisfaction 1.560e-01 3.723e-02 4.190 2.79e-05 ***
For.R$MonthlyIncome 4.129e-05 1.588e-05 2.601 0.009304 **
For.R$PercentSalaryHike 1.960e-02 1.803e-02 1.087 0.276989
For.R$PerformanceRating -1.704e-01 1.823e-01 -0.934 0.350083
For.R$TotalWorkingYears 6.933e-03 1.213e-02 0.572 0.567517
For.R$TrainingTimesLastYear 8.766e-02 3.287e-02 2.667 0.007657 **
For.R$YearsAtCompany -2.839e-02 1.601e-02 -1.773 0.076237 .
For.R$YearsInCurrentRole 7.026e-02 2.045e-02 3.436 0.000591 ***
For.R$YearsSinceLastPromotion -7.068e-02 1.852e-02 -3.816 0.000135 ***
For.R$YearsWithCurrManager 6.796e-02 2.067e-02 3.288 0.001008 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 1298.6 on 1469 degrees of freedom


Residual deviance: 1175.3 on 1457 degrees of freedom
AIC: 1201.3
Number of Fisher Scoring iterations: 5

> probitmodel2 = glm(For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +


For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager, family = binomial(link
= "probit"))
> summary(probitmodel2)

Call:
glm(formula = For.R$Attrition ~ For.R$Age + For.R$JobSatisfaction +
For.R$MonthlyIncome + For.R$TrainingTimesLastYear + For.R$YearsInCurrentRole +
For.R$YearsSinceLastPromotion + For.R$YearsWithCurrManager,
family = binomial(link = "probit"))

Deviance Residuals:
Min 1Q Median 3Q Max
-3.2550 0.2755 0.4849 0.6603 1.1435

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -6.874e-01 2.236e-01 -3.075 0.00211 **
For.R$Age 1.629e-02 5.287e-03 3.081 0.00206 **
For.R$JobSatisfaction 1.504e-01 3.698e-02 4.067 4.77e-05 ***
For.R$MonthlyIncome 3.921e-05 1.254e-05 3.127 0.00177 **
For.R$TrainingTimesLastYear 8.561e-02 3.258e-02 2.628 0.00859 **
For.R$YearsInCurrentRole 5.646e-02 1.921e-02 2.939 0.00330 **
For.R$YearsSinceLastPromotion -8.139e-02 1.752e-02 -4.646 3.38e-06 ***
For.R$YearsWithCurrManager 5.304e-02 1.876e-02 2.828 0.00468 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 1298.6 on 1469 degrees of freedom


Residual deviance: 1180.4 on 1462 degrees of freedom
AIC: 1196.4

Number of Fisher Scoring iterations: 5

You might also like