Utilizing CNN2D for regressor analysis and time series forecasting & modelling of the foods demand supply chain

Abstract

Due to the perishable nature of many food goods and the high stakes involved in mismanaging inventories, precise demand forecasting has assumed paramount importance in recent years. When dealing with data that changes over time, a number of ML and DL methods have lately shown significant gains. Based on an analysis of the Genpact 'Food Demand Forecasting' dataset, this article determines which elements have the most impact on demand, which characteristics are most likely to have an impact, and then suggests an examination of seven regression coefficients to predict future order quantities. We relied on a number of different regression models in our investigation, including Random Forest, Gradient Boosting Regressor, Light Gradient Boosting Machine, Extreme Gradient Boosting Regressor, Cat Boost Regressor, Long Short-Term Memory, and Bidirectional LSTM. This study proves that LSTM is the best algorithm out there and shows how deep learning models may improve predicting.