HR Analytics: Employee Retention at Salifort Motors

Goal: Predict and reduce employee churn by uncovering drivers of turnover through rigorous data analysis and visualization.

Contents:

• data/   Raw & cleaned datasets
• notebooks/ Exploratory analysis & modeling in Python
• tableau/  Interactive dashboards
• presentations/ Final slide deck

---

Requirements

• Company: Salifort Motors
• Challenge: High employee turnover
• Goal: Predict employee churn and identify contributing factors
• Business Impact: Reduce hiring costs and improve retention

Key Skills Demonstrated

• Data cleaning & feature engineering
• Statistical outlier detection (IQR, Z-score)
• EDA with Matplotlib/Seaborn & Tableau
• Predictive modeling (logistic regression, tree-based)
• Dashboard design & storytelling
• Version control & project organization

Data

This folder contains the raw and processed employee data.

Structure

• raw/hr_analytics_cleaned.csv
  Original cleaned export from HRIS. Datasets generated by ETL scripts:
  • `hr_analytics_cleaned.csv'

Notebooks

All exploratory data analysis and modeling lives here.

• hr_analysis.ipynb
  • Data loading, cleaning, outlier detection
  • EDA: satisfaction vs hours/projects/tenure
  • Correlation analysis & feature summaries
  • Initial predictive model for churn

Outlier Detection Methods Explained

IQR Method (Box Plot Method)

• Analogy: Like finding shirts that are unusually small or large in a clothing store
• How it works:
  • Computes Q1 (25th percentile) and Q3 (75th percentile)
  • Defines the “normal range” as Q1 – 1.5×IQR to Q3 + 1.5×IQR (IQR = Q3 – Q1)
• Flagging outliers: Any value below Q1 – 1.5×IQR or above Q3 + 1.5×IQR is flagged as unusual

Z-score Method

• Analogy: Like comparing test scores to the class average
• How it works:
  • Calculates Z-score = (value – mean) / standard deviation
• Flagging outliers: Any value with |Z-score| > 3 (more than 3 SDs from the mean) is considered unusual

---

Metrics & Key Findings

Metric	Key Findings & Business Impact
Satisfaction Level	- Moderate overall mood (0.63/1.0) - Some very unhappy employees (as low as 0.09) - Potential risk for turnover
Last Evaluation	- Generally positive reviews (0.72/1.0) - No extremely poor performers - Consistent evaluation system
Project Count	- Most employees handle 3–4 projects - A few manage 7 projects - Burnout risk at higher project loads
Monthly Hours	- Standard ~40 hours/week on average - Some working up to 77.5 hours/week - Work–life balance concerns
Tenure	- Many outliers (extremely long or short stays) - High variation in stay duration - Retention challenges evident
Work Accidents	- 15% accident rate - Safety concerns - Need for stronger safety protocols
Left Company	- 17% turnover rate (above industry avg.) - Immediate attention needed to reduce churn
Promotions	- Very low promotion rate (2%) - Career growth concerns - Potential driver of turnover

Comprehensive Analysis of HR Visualizations

Turnover Distribution (Basic Bar Chart)

• 17% turnover rate
• Key Risk: High turnover cost impact

Project Load vs Hours (Box Plot & Histogram)

• Optimal project load: 3–4 projects
• Critical finding: 100% turnover for employees handling 7 projects
• Insight: Identified workload threshold beyond which churn spikes

Monthly Hours vs Satisfaction (Scatter Plot)

• High hours (240–315/month): Near-zero satisfaction
• Sweet spot: 210–280 hours/month with satisfaction scores of 0.7–0.9
• Implication: Cultural pressure impacts employee morale

Satisfaction by Tenure (Box Plot & Histogram)

• Two distinct turnover groups:
  • Short-tenure + low satisfaction
  • Medium-tenure + high satisfaction
• Critical tenure mark: Around 4 years

Salary Distribution by Tenure

• No clear correlation between tenure and salary
• Limited salary progression over time
• Retention challenge: Compensation not keeping pace with experience

Monthly Hours vs Last Evaluation (Scatter Plot)

• Overworked high performers are more likely to leave
• Positive correlation between hours worked and evaluation scores
• Highlight: Work–life balance issues driving churn

Promotions & Department Analysis

• Very low promotion rate: ~2% company-wide
• Consistent turnover across all departments
• Conclusion: Retention issue spans the entire organization

---

Key Correlations

• Hours vs Projects: r = 0.42 (strong positive)
• Satisfaction vs Left: r = –0.39 (strong negative)
• Evaluation vs Projects: r = 0.35 (moderate positive)

6. Business Insights

Key Findings

Workload Issues

• 7+ projects ⇒ 100% turnover
• Optimal load: 3–4 projects
• Overworking is prevalent

Satisfaction Patterns

• Two turnover groups:
  • Short tenure + low satisfaction
  • Medium tenure + high satisfaction
• Critical tenure mark identified at ~4 years

Performance Relations

• Positive correlation between hours worked and evaluation scores
• Very low promotion rate (~2%)
• Consistent turnover across all departments

Retention Risks

• Overall turnover rate: 17%
• Work–life balance concerns
• Limited career progression opportunities

Recommendations

1. Implement project caps to prevent overload
2. Review 4-year policies and introduce mid-tenure engagement programs
3. Create clear career paths with promotion roadmaps
4. Address workload culture through flexible scheduling and resource planning
5. Develop a comprehensive retention strategy, including mentorship and recognition initiatives