Building a neural network model for diagnosing the probability of bankruptcy of innovative-active enterprises and checking its adequacy
Abstract
The article is devoted to the substantiation of the choice of financial indicators for discriminant and neural network models for diagnosing the financial condition of innovative active enterprises and determining the probability of their bankruptcy, as well as the construction of these models based on a study of the financial condition of 36 enterprises. The modern imperative of the successful development of the domestic economy is its transition to the rails of innovative development. This process is impossible without competent distribution of financial resources by business entities. In this regard, especially important is the question regarding the development of new approaches and methods for the assessment of readiness of enterprises for implementation of innovation activities due to which investors or, indeed, the state itself will be able to determine the amount of financial resources which is necessary for the development and implementation of new technologies, products or services. It is shown the importance of researching the financial condition of Ukrainian enterprises that are engaged in innovations, since their innovative activity is almost entirely financed by own means. With the aid of Deductor analytical platform, a discriminant model for assessing the financial situation and the probability of bankruptcy for innovative enterprises was built. The neural network model, which together with the analysis "if-then" gives an adequate forecast of the financial state of enterprises engaged in innovation activity, was substantiated and built. Five financial ratios (X1, X2, X3, X4 and X5) are selected and calculated for the analysis of the financial condition of 36 enterprises. For all the studied enterprises (both bankrupt and those against which bankruptcy proceedings were not initiated), the satisfactory forecast was for 30 out of 36 enterprises (83.33%), unsatisfactory for 2 enterprises (5.56%), in the gray zone there were 4 enterprises (11.11%). It is shown that the built neural network model provides forecasts of the financial condition of enterprises and the probability of their bankruptcy at a level significantly higher than discriminant models. The neural network model takes into account the specifics of domestic economic activity of enterprises, because it is built on the basis of financial data of Ukrainian enterprises.
Keywords
