Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

📊 Cogentix Data Analysis Report: High-Performance Workforce Insights

🎯 Executive Summary

Cogentix, a global organization with 10,000+ employees, faces critical challenges in talent retention, promotion fairness, and compensation equity. This report leverages advanced data analytics and machine learning to identify at-risk segments, predict promotion readiness, and flag compensation inequities, providing a data-driven roadmap for HR leadership.


🔍 Key Performance Insights

📉 1. Workforce Engagement & Retention

Analysis of bottom-quartile engagement scores reveals significant demographic variation. Mid-career employees (Age 40-44) exhibit the highest risk for low engagement (27.0%), while gender and location show no statistically significant impact (p > 0.05).

Strategic Takeaway: Retention efforts should shift from broad demographic targeting to role-specific and age-cohort interventions, focusing on manager quality and career progression for the 40-44 age bracket.


🚀 2. Promotion Readiness & Predictive Modeling

Using a leakage-free LightGBM model, we achieved the ability to identify the Top 50 high-potential candidates with a focus on engagement and commitment.

Core Drivers of Promotion:

  • Engagement Score: Primary indicator of readiness.
  • Investment in Learning: Training hours highly correlate with promotion potential.
  • Tenure: Years at company indicate organizational stability.

🏆 3. Performance Drivers & Correlation Analysis

Performance (Rating ≥ 4) is driven more by investment in skills than by department or role. While location-specific variations are significant (p=0.0093), identifying top performers requires looking at individual growth metrics.

Visualizing Performance Drivers:

Performance Correlations Figure 1: Spearman correlations showing training hours and engagement as positive performance drivers.

Departmental Consistency:

Performance by Department Figure 2: Performance remains remarkably consistent across all 10 departments, suggesting a unified organizational standard.

Targeting Skill Growth:

Training Heatmap Figure 3: Heatmap identifying the "Sweet Spot" where training hours maximize performance for mid-tenure employees.


💸 4. Compensation Equity & High-Performer Retention

Our residual-based compensation model identified High-Performing Underpaid employees. By predicting expected salary based on role, performance, and tenure, we flagged individuals whose actual pay falls below the 20th percentile of expectation.

Action Item: Review flagged employees in the outputs/csv_reports/underpaid_high_performers.csv to preemptively address turnover risk among top talent.


🚨 Leadership Accountability: Manager Red Flags

Statistically significant low engagement was detected in 27 manager-led teams. These teams show average performance ratings lower than the company average, indicating a direct link between leadership quality and workforce output.

Highlighted Metric Value
Managers Flagged 27
Highest Risk Team Size 22
Engagement p-value < 0.001

🗺️ Strategic Roadmap

🔴 Immediate (0-3 Months)

  • Salary Adjustments: Address gaps for flagged underpaid high performers.
  • Leadership Coaching: Intervene with the 27 flagged managers.
  • Promotion Reviews: Evaluate the Top 50 candidates identified by the predictive model.

🟡 Strategic (3-12 Months)

  • Training Expansion: Incentivize high-impact training hours (60+ hours/year).
  • Age-Cohort Programs: Launch engagement initiatives for the 40-44 age group.

🟢 Long-Term (12+ Months)

  • Predictive Integration: Incorporate promotion models into standard HR reviews.
  • Market Benchmarking: Review external competitiveness for niche roles identified in equity analysis.

🛠️ Technical Appendix

  • Stack: Python (Pandas/NumPy), LightGBM, Scikit-learn, Plotly, Streamlit.
  • Rigors: Chi-Square, Kruskal-Wallis, Spearman Correlation, Wilson Confidence Intervals.
  • Project Structure: Organized into data/, notebooks/, app/, and outputs/.

Last Updated: February 2026 | Analysis by VRMithun

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages