Application Of Data Mining Techniques For Understanding Capital Structure Of Brazilian Companies
Free (open access)
103 - 111
R. A. M. Horta, H. M. Pires & B. S. L. P. de Lima
The capital structure decision is related to the adoption of strategies for the choice of financing sources for the own capital and the thirty party capital. Numerous studies were developed aiming at clarifying the adoption of those strategies. This paper aims to identify attributes that influence the decision of capital structure through an exploratory data analysis using records of Brazilian companies of open capital. The financial companies were not included in this study. Thus, it is expected to achieve new information on this process of knowledge discovery through statistical and data mining techniques, such as linear regression, robust regression and neural networks, in order to get a better assessment of the financing strategies employed by companies. The data analysis was made in two stages employing as dependent variables the onerous debt and the equity degree of financial leverage. The database consists of companies on BOVESPA classified by economic sectors in the period from 1996 to 2006. The results suggest that the explanatory variables of financial leverage are: economic sector; return on equity; return on assets; immediate liquidity; general liquidity and assets turnover. Keywords: data mining, capital structure, attribute selection.
data mining, capital structure, attribute selection.