PREREQUISITES FOR THE USE OF MACHINE LEARNING FOR BUSINESS VALUATION

Resumo

Goal: The paper examines the fundamental theoretical prerequisites for the use of machine learning in business valuation. Methods: The study demonstrates that the use of statistical methods addresses the shortcomings of traditional approaches to valuation, in particular, the income approach and the discounted cash flow method. Results: Substantiation is given for the rejection of traditional econometric methods (linear regression, estimated by the least squares method) in favor of more complex nonparametric statistical models. Conclusion: Machine learning expands the empirical toolkit of the economist, allows for small datasets, solves the problem of asset valuation complexity, protects against false discoveries, and does not require compliance with Gauss-Markov assumptions. The paper also addresses the black box problem "“ the difficulty of interpreting models derived from statistical learning.