Aplikace neuronových sítí při stanovení generátorů hodnoty podniku

In general, the primary strategic objective of businesses is the growth of their value, the maximization of their wealth for shareholders. Mainly the complexity of today´s businesses and the growing customer demands are the reasons why business value depends on a whole range of different aspects. Business value must be effectively at all levels. The means of value management are the so-called business value generators, and their identification is a critical task of each business management. This dissertation uses an innovative approach of artificial neural networks to identify value generators in the manufacturing industry. The data for the research was obtained from the CRIBIS database, specifically the financial statements of processing companies active in the years 2015-2019. After the necessary adjustments of the data and defining possible value generators in the form of production factors, financial analysis and analysis of bankruptcy and creditworthiness models of the median manufacturing company in the Czech Republic was used in order to analyse the basic dataset and add complexity. Subsequently, cluster analysis using Kohonen networks was carried out, which divides the companies into clusters, where the value generators are the searched for. Neural networks are also used for the identification of value generators by means of regression analysis in three experiments, which differ in used continuous and categorical predictors. A positive correlation was identified between the performance of a company operating in the manufacturing industry and reinvestment. Furthermore, based on the results of the financial analysis, it can be stated that in the monitored period, the median company does not show any engagement with the concept of Industry 4.0. According to bankruptcy and creditworthiness models, this company is rather in the grey zone, as clear creditworthiness has not been proven. The results of the experiments indicate that neural networks are a suitable means to identify value generators; however, categorical predictor or division of companies into clusters using cluster analysis does not improve the performance of the networks. The network 2. MLP 7-9-1 with its performance of nearly 90 % is selected as the most successful. What is crucial is the finding that all items representing production factors are very important for the value of processing companies operating in the Czech Republic, as measured by EVA Equity. Depreciation of fixed intangible and tangible assets, material and energy consumption, and personnel costs are of great importance.