Application of different Machine Learning models for Supply Chain Demand Forecasting: Comparative Analysis

Abstract

Demand is defined as the propensity or willingness of customers to pay a certain amount of price for a product or service they desire. Business entities use various forecasting techniques to anticipate customer demands in advance to make crucial strategic decisions related to various aspects of the supply chain, such as customer service level, inventory management, manufacturing planning, and control, etc. But the error related to forecasting models creates uncertainty and poses a great challenge to decision-makers. With the ever-changing market dynamics, it is becoming more and more important for businesses to minimize those forecasting errors. In this paper, we have analysed the application of various advanced Machine Learning algorithms like Multi-layered Perceptron model (MLP), Convolution Neural Network (CNN), Long-Short Term Memory (LSTM) Networks, etc. in Time Series Forecasting, thereafter, performed a comparative analysis to understand which of them yields the better result. To perform the analysis, we have considered the ‘Store Item Demand Forecasting’ dataset available at Kaggle.