Measuring Factors of Employment by Classification Tree Models
A. Nachev, BIS, Cairnes Business School, NUI Galway, Galway, Ireland.
Manuscript received on 10 October 2017 | Revised Manuscript received on 18 October 2017 | Manuscript Published on 30 October 2017 | PP: 22-30 | Volume-7 Issue-1, October 2017 | Retrieval Number: A5174107117/17©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This paper presents a case study on data mining modeling, based on classification trees. The study analyzes data from a national household survey, which provides information about Irish labour and unemployment status of the respondents. Based on trained predictive models, we address some gaps in previous studies by providing means to measure and rank the employment factors and analyze their role over the studied period. Results from experiments show that features representing age and education appear as top factors affecting the employment status. Studying further each of those by VEC analysis, we find empirically the role of their values in employment success. Measuring the model performance, we came to the conclusion, that a carefully trained classification tree can outperform neural networks trained on the same data in terms of accuracy, but underperforms neural nets in terms of AUC.
Keywords: Classification, Data Mining, Labour, Classification Trees.
Scope of the Article: Classification