IntelliAuto Employee Retention Analysis

Executive Summary
This report presents a comprehensive analysis of employee retention for IntelliAuto, an
automobile parts manufacturer planning to transition to a digital manufacturing factory
over the next 8 years. Using logistic regression analysis, we identified employees with a
high probability of leaving the company within 8 years and developed prescriptive
models to optimize incentive allocation strategies.

Our analysis revealed that education years, union membership, work experience, and
promotion history are significant predictors of employee retention. We identified 10
employees with the highest probability of leaving, primarily from Professional,
Management, and Tech/Sales departments.

Two incentive allocation strategies were modeled: 1. A 20% probability reduction

strategy requiring an investment of 59.32K, optimally allocating 17.57 seminar hours and
40K in salary increases 2. A 60% probability reduction strategy requiring a significantly
higher investment of 172.70K, exclusively through salary increases

We recommend implementing the 20% reduction strategy as it provides a more

balanced and cost-effective approach to retention. For long-term retention, IntelliAuto
should focus on promotion opportunities, union membership benefits, and targeted
professional development for employees with higher education levels.

Introduction
IntelliAuto, an automobile parts manufacturer with 5000 employees, is planning to
transition to a digital manufacturing factory over the next 8 years. This digital
transformation, while beneficial for productivity and competitiveness, raises concerns
about workforce disruption and employee retention. Understanding factors that
influence employee turnover is crucial for successful transition planning.

This analysis aims to: 1. Identify factors that predict employee retention 2. Determine
employees with high probability of leaving within 8 years 3. Develop optimal incentive
allocation strategies to improve retention 4. Provide recommendations for workforce
management during digital transformation
The analysis uses data from a survey of 822 randomly selected IntelliAuto employees,
applying logistic regression and prescriptive analytics to develop evidence-based
retention strategies.

Problem Formulation

Logistic Regression Model

To identify employees with high probability of leaving within 8 years, we developed a

logistic regression model:

$$P(Leave < 8 years) = \frac{1}{1 + e^{-z}}$$

Where: $$z = \beta_0 + \beta_1 \times WorkHrs + \beta_2 \times Age + \beta_3 \times
Educ_Yrs + \beta_4 \times Sex + \beta_5 \times MemUnion + \beta_6 \times WrkYears +
\beta_7 \times NumPromo + \beta_8 \times FutPromo + \beta_9 \times AwareI4$$

The model estimates the probability of an employee leaving within 8 years based on
various demographic and employment factors.

Prescriptive Optimization Model

For the incentive allocation strategy, we formulated a linear programming model:

Objective Function: Minimize Total Cost = ∑(Seminar Hours × Seminar Cost + Salary
Increase)

Subject to constraints: 1. Total seminar hours ≤ 20 2. Total salary increase ≤ 40K 3. For
each employee i: Probability Reduction ≥ Required Reduction Target

Where: Probability Reduction = Seminar Hours × Seminar Effect + Salary Increase ×

Salary Effect

Parameters: - Seminar Effect by Department: - Management: 0.01 per hour -

Professional: 0.02 per hour - Tech/Sales: 0.03 per hour - Admin: 0.01 per hour - Service:
0.03 per hour - Production: 0.03 per hour - Labourer: 0.02 per hour

• Seminar Cost by Department:

• Management: 0.8K per hour
• Professional: 1.0K per hour
• Tech/Sales: 1.1K per hour
• Admin: 0.9K per hour
• Service: 1.2K per hour
 • Production: 1.1K per hour

• Labourer: 0.9K per hour

• Salary Effect by Gender:

• Male: 0.05 probability reduction per 1K

• Female: 0.03 probability reduction per 1K

Two scenarios were modeled: 1. 20% probability reduction target 2. 60% probability
reduction target

Findings and Discussions

Logistic Regression Analysis

The logistic regression model identified several significant predictors of employee

retention:

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 1.619901 0.895446 1.809 0.07044 .
WorkHrs 0.001117 0.009938 0.112 0.91051
Age -0.030659 0.019983 -1.534 0.12498
Educ_Yrs 0.198119 0.038563 5.138 2.78e-07 ***
SexMale -0.021121 0.216056 -0.098 0.92213
MemUnionYes Union -1.444976 0.265949 -5.433 5.53e-08 ***
WrkYears -0.065207 0.020768 -3.140 0.00169 **
NumPromo -0.792813 0.082144 -9.651 < 2e-16 ***
FutPromo2.Likely 0.401805 0.289979 1.386 0.16586
FutPromo3.Not sure -1.012431 0.473308 -2.139 0.03243 *
FutPromo4.Unlikely 0.131369 0.296712 0.443 0.65795
FutPromo5.V Unlikely -1.099607 0.277637 -3.961 7.48e-05 ***
AwareI4Yes 0.038492 0.189745 0.203 0.83924

Key findings: 1. Education Years: Higher education significantly increases the

probability of leaving (p < 0.001) 2. Union Membership: Union members are significantly
less likely to leave (p < 0.001) 3. Work Experience: More years of work experience
correlates with lower turnover (p < 0.01) 4. Number of Promotions: More promotions
significantly reduces turnover risk (p < 0.001) 5. Future Promotion Prospects:
Employees uncertain or very unlikely to be promoted show lower turnover risk

Interestingly, work hours, age, gender, and Industry 4.0 awareness did not significantly
predict turnover.
Top 10 Employees at Risk of Leaving

The model identified the following employees with highest probability of leaving within
8 years:

Employee ID Department Gender Age Work Hours Probability

51 Professional Male 31 40 0.984

173 Management Male 24 50 0.983

56 Management Male 25 40 0.981

35 Tech/Sales Female 27 55 0.981

517 Tech/Sales Female 27 55 0.981

256 Professional Female 31 45 0.981

679 Professional Female 31 45 0.981

727 Professional Female 31 45 0.981

808 Professional Female 31 45 0.981

761 Tech/Sales Female 27 47 0.981

Notable patterns: - Professional department has the highest representation (5

employees) - Tech/Sales department has 3 employees at high risk - Management
department has 2 employees at high risk - All employees are relatively young (24-31
years old) - All have high probabilities (>98%) of leaving within 8 years

Prescriptive Analytics Results

20% Probability Reduction Strategy

The optimal solution for reducing turnover probability by 20% requires: - Total cost:
59.32K - Seminar hours allocation: 17.57 hours - Salary increases: 40K (full budget
utilized)

Resource allocation by employee:

 Employee Seminar Salary Increase Total Cost
Department Gender
ID Hours (K) (K)

51 Professional Male 0.00 3.94 3.94

173 Management Male 0.00 3.93 3.93

56 Management Male 0.00 3.92 3.92

35 Tech/Sales Female 0.00 6.54 6.54

517 Tech/Sales Female 0.00 6.54 6.54

256 Professional Female 0.00 6.54 6.54

679 Professional Female 0.00 6.54 6.54

727 Professional Female 8.55 0.00 8.55

808 Professional Female 9.02 0.00 9.02

761 Tech/Sales Female 0.00 2.05 6.99

This strategy balances seminar hours and salary increases, focusing resources where
they provide the most cost-effective probability reduction.

60% Probability Reduction Strategy

The standard model with original constraints (20 seminar hours, 40K salary budget)
could not achieve a 60% probability reduction for all 10 employees. A relaxed model
without these constraints yielded:

• Total cost: 172.70K

• Seminar hours allocation: 0 hours
• Salary increases: 172.70K

Resource allocation by employee:

Employee Seminar Salary Increase Total Cost

Department Gender
ID Hours (K) (K)

51 Professional Male 0.00 11.81 11.81

173 Management Male 0.00 11.79 11.79

56 Management Male 0.00 11.77 11.77

 Employee Seminar Salary Increase Total Cost
Department Gender
ID Hours (K) (K)

35 Tech/Sales Female 0.00 19.62 19.62

517 Tech/Sales Female 0.00 19.62 19.62

256 Professional Female 0.00 19.62 19.62

679 Professional Female 0.00 19.62 19.62

727 Professional Female 0.00 19.62 19.62

808 Professional Female 0.00 19.62 19.62

761 Tech/Sales Female 0.00 19.62 19.62

This strategy requires a significantly higher budget and relies exclusively on salary
increases, as they provide more efficient probability reduction per dollar invested
compared to seminars.

Comparative Analysis

The 20% reduction strategy offers a more balanced and cost-effective approach: -
Utilizes both incentive types (seminars and salary increases) - Requires less than 35% of
the budget needed for the 60% reduction - Provides a reasonable probability reduction
with available resources

The 60% reduction strategy: - Requires more than 4 times the original salary budget -
Relies exclusively on salary increases - May not be financially feasible within current
constraints

Conclusion and Recommendations

Key Insights

1. Retention Risk Factors: Education level, lack of union membership, limited work
experience, and fewer promotions are the strongest predictors of turnover risk.

2. Department Distribution: Professional, Tech/Sales, and Management

departments have the highest concentration of at-risk employees, suggesting these
areas may be most vulnerable during digital transformation.

3. Incentive Effectiveness: For the identified high-risk employees:

 4. Salary increases are more cost-effective for male employees
5. Seminar hours are more cost-effective for female employees in Professional roles
6. A balanced approach utilizing both incentives optimizes resource allocation

Strategic Recommendations

1. Implement the 20% Reduction Strategy:

2. Allocate 17.57 seminar hours and 40K in salary increases as detailed in the
prescriptive model
3. Focus seminar hours on female Professional employees where they provide the
highest return

4. Distribute salary increases across other employees based on the optimization

model

5. Enhance Promotion Opportunities:

6. Develop clear career advancement paths, especially for highly educated employees
7. Implement regular promotion reviews and transparent advancement criteria

8. Create new leadership roles related to digital transformation initiatives

9. Strengthen Union Relationships:

10. Collaborate with unions during digital transformation planning

11. Highlight union benefits in employee communications

12. Consider extending union-like benefits to non-union employees in high-risk

departments

13. Targeted Development Programs:

14. Create specialized professional development for employees with higher education
15. Design training programs that align with Industry 4.0 skill requirements

16. Establish mentorship programs pairing experienced employees with newer staff

17. Long-term Digital Transformation Strategy:

18. Develop a phased implementation plan with clear communication at each stage
19. Create cross-functional teams including representatives from high-risk
departments
20. Establish regular feedback mechanisms to address concerns proactively
By implementing these recommendations, IntelliAuto can mitigate turnover risk during
its digital transformation while optimizing resource allocation for maximum retention
impact.

References
1. IntelliAuto Employee Satisfaction Survey Dataset
2. Industry 4.0 workforce impact studies
3. Logistic regression and prescriptive analytics methodologies

Appendix
The complete R code, Python optimization model, and additional visualizations are
available in the accompanying files.