Maximizing Financial Management efficiency with a novel Machine Learning algorithm

Abstract

Any firm must successfully manage its financial assets to succeed. To make wise choices for handling resources, possibilities for investment, and allocation of funds, accountants rely on reliable forecasting of finances. To enhance the effectiveness of Financial Management (FM), this research develops hybrid long short-term memory and hierarchical analytic process (HLSTM-AHP) technique. The LSTM approach is applied to develop the financial evaluation system, while the AHP approach is employed to establish the weightings of economic variables incorporated into the LSTM framework. To show how well the suggested HLSTM-AHP approach works at enhancing FM effectiveness, real-time accounting information from a company that is publicly traded are implemented. To further address the issue of anomalous data regarding finances, this research employs a unique sampling data collected by linear discriminant analysis (LDA) to develop a multiscale convolutional neural network (MCNN), which improves the framework's forecasting performance and demonstrates conclusively that machine earning (ML) is practicable in the study of FM forecasting, with plenty of opportunity for future investigations.