Forecasting financial indicators of the film production market using artificial intelligence

The subject of the research is the problem of forecasting financial indicators for service-oriented markets under extraordinary conditions. The goal of the work is to develop an effective approach for forecasting the financial indicators of the film production market, which is based on recurrent and convolutional neural networks and uses natural language processing tools to convert descriptive data into numerical form. The article addresses the following tasks: defining a set of indicators capable of describing the state of the film production market from the point of view of the company, the external environment, and the target audience; development of algorithms for processing numerical and textual information; determining the list of target neural networks and revealing the specifics of their implementation; determination of the most effective prognostic approach by solving the problem of linear optimization. The following methods used are – an analytical method for determining a set of neural networks; expert assessment for the formation of the most important independent indicators and determination of efficiency factors; experimental, multi-criteria evaluation to determine the most effective model. The following results were obtained: a set of data reprocessing algorithms was formed for their longer use in recurrent and convolutional neural networks. Several common architectures involving MapReduce technology have been implemented. It was determined that the most effective model is a bidirectional recursive neural network with support for long- and short-term memory. The expediency of using parallelization technology is shown and a set of open questions for further research is defined. Conclusions: conducting an analysis of algorithms for forecasting financial indicators based on artificial intelligence with subsequent experimental verification made it possible to form a relatively effective way of predicting the state of indicators of the film industry market under extraordinary conditions. The obtained results allow us to affirm the feasibility of implementing the proposed approach, which can influence the policy formation of the film producer's company or the fund that operates the company's financial instruments. At the same time, there are ways to further improve the results with the involvement of alternative approaches both to parallelization and to forecasting in general.