Neural Network Model and Decision Support Methodology Based on Machine Learning for Time Series Forecasting

Abstract

This article considers the issue of building an optimized neural network model for time series forecasting. In this case, the training sample of size  using machine learning methods is based on the solution of the correlation-regression problem for training the training sample. The research is based on the analysis of tax authority data. In this case, a neural network-based model and a decision-making method are proposed based on the analysis of tax revenue types and states. In addition, an algorithm for intelligentizing decision-making is proposed.