Authors: Professor Dr. Rajendra Singh Kushwah, Ritesh Kumar
Abstract: The most popular method of forecasting SPs throughout the course of history was to make use of statistical models such as moving averages, ARIMA, and the GARCH. This was the most common strategy. It is difficult for these models to accurately predict future SPs due to the fact that financial markets are characterized by their complexity, non-linearity, and unending change. In spite of the fact that these models are able to effectively recognize particular linear trends within historical data, they often have a difficult time accurately predicting events that will occur in the future. As a result of the fact that price changes are impacted by a large variety of different factors, these models do not allow for the precise estimation of future SPs. These models are not particularly accurate as a consequence of this. In recent years, there has been an increase in interest in the topic of financial forecasting in relation to the capabilities of machine learning and deep learning approaches. This attention has brought about a number of interesting developments. The methodologies in question are able to accurately characterize not only the time-dependent patterns that are present in the data, but also the intricate and non-linear linkages that are there. In the field of time series prediction, recurrent neural networks (RNN) and long short-term memory (LSTM) networks, in addition to generalized recurrent units (GRU), have been shown to be successful tools. When it comes to capturing long-range dependencies in supply chains, it is preferable to traditional methods since it is more accurate.
