LinkedIn-Based Models Predict Indian IT Tenure Within 12.4 Months Across 411 Job Spells
Updated
Updated · BIOENGINEER.ORG · Nov 14
LinkedIn-Based Models Predict Indian IT Tenure Within 12.4 Months Across 411 Job Spells
1 articles · Updated · BIOENGINEER.ORG · Nov 14
Summary
Researchers used public LinkedIn profiles to estimate how long Indian IT professionals stay in a job, turning 411 completed employment spells into a continuous tenure-prediction dataset.
Decision tree regressors delivered the best test accuracy with a 12.3899-month RMSE, while Neural networks also performed strongly and consistently across training and test data.
Experience and organizational seniority emerged as the strongest predictors of longer tenure, while certifications also helped; larger LinkedIn networks pointed the other way, correlating with shorter stays.
Three clusters sharpened that pattern: experienced professionals averaged 47.1 months in a role, freshers 24.3 months, and a small group of highly connected "influential freshers" just 17.73 months.
The authors say firms could use the forecasts for staffing, promotions and targeted upskilling, but only as post-hire advisory tools rather than automated hiring filters.