Prophet-random forest-based e-commerce event-driven demand prediction method

Abstract

The invention discloses an e-commerce event-driven demand prediction method based on Prophet-random forest. The method comprises the steps: obtaining the historical sales data of an e-commerce platform, and the sales data comprises the time sequence data and the user data of purchasing related products; cleaning historical sales data to improve data quality; performing scale compression on the time series data to reduce data volatility; performing event modeling based on festival activities and e-commerce platform promotion events; adding an event regression amount to the Prophet model, carrying out Prophet prediction, removing and interpolating outliers belonging to an event effect range, and carrying out random forest prediction; prophet prediction and random forest prediction results are combined; and precision evaluation is carried out to observe the generalization ability and the prediction effect of the model. According to the invention, prediction is carried out by using Prophetand a random forest according to the characteristics of an e-commerce event demand and a non-event demand, so that the prediction precision of an e-commerce event-driven demand is improved.

Description

technical field [0001] The invention belongs to the field of e-commerce demand forecasting, and in particular relates to an e-commerce event-driven demand forecasting method based on Prophet-Random Forest. Background technique [0002] The date interval between festival activities and e-commerce platform promotion events is not fixed, and does not show a standard periodic movement law. The short-term consumer demand driven by it shows a short-term surge in events and a small fluctuation in the period affected by non-event effects. state. However, the current commonly used method for e-commerce demand forecasting is the autoregressive moving average model. The premise of its time-series data forecast is that the data is stable, and it cannot capture the law of unstable data well. When the platform promotion event is driven and has sudden characteristics, the existing model cannot represent the demand in the real business environment. Contents of the invention [0003] The...

AI Technical Summary

Problems solved by technology

However, the current commonly used method for e-commerce demand forecasting is the autoregressive moving average model. The premise of its time-series data forecast is that the data is stable, and it cannot capture the law of unstable data well. When the platform promotion event is driven and has sudden characteristics, the existing model cannot reflect the demand in the real business environment

Images