Forecasting and Inventory Planning for E-Commerce using Advanced Time Series Models

Abstract

The use of advanced time series models, including SARIMA, Prophet, and LSTM, in forecasting demand for improved inventory planning is discussed in this paper. LSTM demonstrates superior performance over SARIMA and Prophet, thus ensuring more accurate forecasts in catching complex patterns and long-term dependencies. Consequently, improved accuracy of the LSTM-based forecast leads to better inventory management in that it reduces stockouts and overstocking. It would cover how the implementation of machine learning algorithms and integration of real-time data contribute to optimizing operations for e-commerce. Future potential improvements include hybrid models as well as the integration of macroeconomic factors.