TY - GEN
T1 - Using multi-layer perceptrons for analysis of labour data
AU - Nachev, A.
N1 - Publisher Copyright:
© 2017 CSREA Press. All rights reserved.
PY - 2017
Y1 - 2017
N2 - This paper presents a case study, which explores factors affecting employment of Irish population, using data from quarterly national household survey containing 199,264 records and 115 variables. After data clensing and variable selection, predictor variables were grouped as demographic, educational, and occupational. We used MLP neural network to buid predictive models that fit data in order to explore interactions between the variables and to provide insight on their role in employment. We estimated variable significance, reduced dimensionality, and selected optimal neural network architecture. The model predictive abilities were esimated by accuracy, ROC analysis, and AUC. In order to provide solid results, we applied rigorous testing procedure combining two testing partitions, cross-vlidation and iterations. Our results show that the MLP model outperforms logistic regression and linear discriminant analysis. We also did VEC analysis of the most significant factors of employment and commented on their characteristics.
AB - This paper presents a case study, which explores factors affecting employment of Irish population, using data from quarterly national household survey containing 199,264 records and 115 variables. After data clensing and variable selection, predictor variables were grouped as demographic, educational, and occupational. We used MLP neural network to buid predictive models that fit data in order to explore interactions between the variables and to provide insight on their role in employment. We estimated variable significance, reduced dimensionality, and selected optimal neural network architecture. The model predictive abilities were esimated by accuracy, ROC analysis, and AUC. In order to provide solid results, we applied rigorous testing procedure combining two testing partitions, cross-vlidation and iterations. Our results show that the MLP model outperforms logistic regression and linear discriminant analysis. We also did VEC analysis of the most significant factors of employment and commented on their characteristics.
KW - Business intelligence
KW - Classification
KW - Labour analysis
KW - Neural networks
UR - https://www.scopus.com/pages/publications/85068211560
M3 - Conference Publication
AN - SCOPUS:85068211560
T3 - 2017 World Congress in Computer Science, Computer Engineering and Applied Computing, CSCE 2017 - Proceedings of the 2017 International Conference on Artificial Intelligence, ICAI 2017
SP - 223
EP - 229
BT - 2017 World Congress in Computer Science, Computer Engineering and Applied Computing, CSCE 2017 - Proceedings of the 2017 International Conference on Artificial Intelligence, ICAI 2017
A2 - Arabnia, Hamid R.
A2 - de la Fuente, David
A2 - Kozerenko, Elena B.
A2 - Olivas, Jose A.
A2 - Tinetti, Fernando G.
PB - CSREA Press
T2 - 2017 International Conference on Artificial Intelligence, ICAI 2017 at 2017 World Congress in Computer Science, Computer Engineering and Applied Computing, CSCE 2017
Y2 - 17 July 2017 through 20 July 2017
ER -