Neural networks have recently been gaining popularity in the business practice. Research has even confirmed their better performance over traditional methods. This paper gives an overview of one of the types of neural networks, generalized regression neural networks. These are then used to establish a plan for the future sales of a company. However, generalized regression neural networks also have their drawbacks. They are oversized and have a long computation time. Despite these disadvantages the article searches for, based on data from the profit and loss accounts of the food company Friall, s.r.o. from the years 1995-2015, the dependence of revenues on production factors. 1000 random neural structures are generated, from which the 5 most appropriate are preserved using the method of least squares. Additionally, a sensitivity analysis is conducted to determine how the individual production factors affect the firm's ability to generate revenue. The proposed neural network is potentially applicable in practice when compiling the financial plan of a company derived from the amount of sales.
Financial planning within a company represents one of the basic activities of a company as well as a very demanding activity of the financial manager. Based on the main principles of economy and data from the previous periods (especially main financial statements – balance sheet, profit and loss, cash-flow statement) the future development of the given company is predicted. Different mathematical and statistical models and methods including statistical, causal and intuitive methods are used to carry this out. Some models, such as artificial neural networks, represent a very efficient method for prediction. The article identifies company revenues as the initial indicator in setting a company´s financial plan. Multilayer perceptron neural networks are very often used for their prediction. Information on Hornbach Company´s Profit and Loss Statement from the period of 1999 to 2015 is used as input data. A total of 1000 artificial neural networks is generated, out of which 5 most appropriate are maintained. Sensitivity analysis is also carried out. The contribution finally states that in practice, the suggested neural structures are useful for compiling a company financial plan which is always derived from the amount of sales.
JavaScript is turned off in your web browser. Turn it on to take full advantage of this site, then refresh the page.