A smart system for short-term price prediction using time series models

Abstract

The primary goal of this paper is to develop a smart system for short-term price prediction for various products using time series models. The system includes a series of processes, e.g., extracting sales data from a website, pre-processing raw data, and using an Autoregressive Integrated Moving Average (ARIMA) model We investigate that traditional ARIMA techniques suffer with performance issues due to identifying the parameter settings therefore, we use auto ARIMA for our project. To evaluate the prediction accuracy of our approach, we use the Root Mean Square Error (RMSE) and the Mean Absolute Percentage Error (MAPE) as performance metrics. We test on two seasonal products while considering different brands of each product. The sales data are taken from the PriceMe website. Furthermore, we also compare the ARIMA model with Moving Average (MA) model. In the case of the MA model, we find that the forecast trends are represented by a flat line. Also, the auto ARIMA model is not appropriate for predicting long-term trends.