Forecasting Amazon’s Quarterly Net Sales Based on Time Series
Shihan Wang · Advances in Economics Management and Political Sciences · 2024
This paper presents an in-depth analysis of the Autoregressive Integrated Moving Average (ARIMA) model for forecasting Amazon’s quarterly net sales, using historical data from 2007 to 2019. The model’s ability to handle trends and seasonality in time series data is highlighted. The study outlines the data transformation process, including log transformations and differencing, to ensure stationarity before model development. Four potential ARIMA models were constructed based on the observed autoregressive and moving average characteristics. The ARIMA(3,1,4) model was ultimately selected for its optimal balance between simplicity and prediction accuracy. A thorough diagnostic assessment was then conducted to ensure that the selected model met important assumptions of the ARIMA framework. The study proceeds to forecast Amazon’s sales for the next eight quarters in 2020 and 2021, demonstrating the model's practical utility in predicting future sales trends. The insights obtained aim to optimize inventory management, improve resource allocation, and better understand seasonal sales fluctuations. These findings offer strategic insights for e-commerce decision-makers.