Where academic tradition
meets the exciting future

A Practical Classification Model for Small and Medium Sized Enterprises According to Their Profitability

Dumitru Iulian Nastac, Irina-Maria Dragan, Alexandru Isaic-Maniu, A Practical Classification Model for Small and Medium Sized Enterprises According to Their Profitability. TUCS Technical Reports 1165, Turku Centre for Computer Science, 2016.

Abstract:

The SMEs sector in Romania, still ongoing consolidation, is particularly vulnerable to the specific factors of the market economy fluctuations, an economy not yet fully functional as a consequence of the transition from planned communist type economy, which had operated nearly half a century. The purpose of the performed analysis is to identify some potential factors for influencing the companies increasing performance, in order to provide several foundation elements for the governmental policies and strategies in the field. The study uses exhaustive balance sheet information, not obtained by sampling, and refers to the strongest SMEs segment, the medium-sized enterprises (n=7,902 units of a total of N=572,800 SMEs with approved financial documents), these being likely to be closer to the functioning company's mechanism. The balance sheet data processing, for the medium-sized enterprises, was performed using an artificial neural networks (ANNs) model which was validated by comparing its outcome with other results provided by multiple statistical regression models. The main goal was to establish a flexible ANN classifier, for these medium-size enterprises, which can be further adapted to eventually nonstationary changes. These first results are encouraging for further developments in order to obtain an adaptive system.

BibTeX entry:

@TECHREPORT{tNaDrIs16a,
  title = {A Practical Classification Model for Small and Medium Sized Enterprises According to Their Profitability},
  author = {Nastac, Dumitru Iulian and Dragan, Irina-Maria and Isaic-Maniu, Alexandru},
  number = {1165},
  series = {TUCS Technical Reports},
  publisher = {Turku Centre for Computer Science},
  year = {2016},
  keywords = {SMEs, neural networks, classification, econometric models, strategic measures, validation tests},
}

Belongs to TUCS Research Unit(s): Computational Biomodeling Laboratory (Combio Lab)

Edit publication