The use of classification techniques applied to the profiling of workers in the National Employment System: a machine learning approach
Machine learning, Public employment system, Worker profiling, Supervised learning
The present work proposes the application of machine learning in the profiling of workers of brazilian public employment system, the National Employment System - NES ( Sine in portuguese ). The use of an automated mechanism of workers profiling will allow efforts to be directed to the preventive treatment of those workers most likely to remain outside the formal labor market longer. Accuracy and F1-Score were used as metrics in the evaluation of the proposed models. The model with the best results in the experiments was the Extreme Gradient Boosting. Future improvements may include the addition of features related to workers profile, to local labor market and to transitions of occupations, economic sectors and residence