Method and device for predicting data

Abstract

The invention provides a method and a device for predicting data and capable of improving the data predicting accuracy. The method includes: step A, optimizing parameters of a plurality of preselected data prediction models respectively; step B, predicting historical data through the plurality of data prediction models after parameter optimization, and determining an optimized data prediction model according to differences between respective predicting results of the plurality of data prediction models and the historical data; step C, predicting future data through the optimized data prediction model; and repeating the step A to the step C sequentially when the difference between the predicted result in the step C and an actual occurrence value of the future data exceeds a preset range.

Description

technical field [0001] The present invention relates to methods and apparatus for forecasting data. Background technique [0002] Sales forecast is the key to effective inventory control for e-commerce retail enterprises. Excessive inventory will cause a backlog of goods, affecting capital turnover and profitability; too small inventory will cause out-of-stock goods, affecting sales and customer experience. However, the environment of the e-commerce retail market is complex and changeable. Factors such as market promotions one after another, shifts in consumer preferences, and seasonal cycle changes in commodities have resulted in the diversity and variability of commodity sales characteristics. The complex and changeable market environment and sales characteristics have brought challenges to the accurate forecasting of commodity sales. [0003] For the sales forecast of a single commodity, the existing technical solutions usually use the historical sales data of the commod...

AI Technical Summary

Problems solved by technology

[0009] In the process of realizing the present invention, the inventor found that any forecasting method is based on certain assumptions. The main shortcoming of the existing sales forecasting technical solution is that when the sales characteristics of the forecasted commodity change, the current Some forecasting models cannot be changed and adjusted automatically to adapt to new situations

[0010] For the weighted moving average method, the weight is not easy to determine, there is no fixed rule to follow, it is too random, and it is difficult to deal with sales with seasonal fluctuation characteristics, so it is difficult to meet the requirements in accuracy; for the exponential smoothing method Said that this method can fit different sales curve characteristics, but it has similar problems with the weighted moving average method - it is difficult to determine the smoothing coefficient, and once the smoothing coefficient is determined, it cannot be changed according to the stage characteristics of sales; for ARIMA For the model prediction method, when the sales characteristics change, the order of difference, autocorrelation coefficient and partial autocorrelation coefficient truncated order may all change, which means that the parameters of the model need to be adjusted accordingly

[0011] However, the complex and changeable market environment of the e-commerce retail industry will inevitably lead to frequent changes in the characteristics of commodity sales. In addition, the number of commodities is often counted in millions. The forecast of commodity sales in the prior art has a large deviation, which often leads to commodity out-of-stock or commodity backlog.

Images