Poisson process user-to-shop behavior prediction based on automatic fitting mean function

Abstract

The invention discloses a Poisson process user-to-shop behavior prediction method based on an automatic fitting mean function, and in the prediction method, the Poisson process modeling prediction is performed by the means of an automatic fitting mean function in a case where a user-to-shop behavior of buying a commodity has a strong randomness, so as to fully analyze whether a user will go to the shop to buy a commodity in the future. According to the invention, the problems of the unknown supply and demand information relationship of the number of shops and commodity demanders in the unknown supply and demand information reflection is solved, and the problem that the customer loyalty cannot be evaluated in part is solved; meanwhile, a set of prediction algorithm suitable for the prediction of user-to-shop purchase and integrated with data acquisition, parameter training, training result analysis and modeling prediction is developed, the most appropriate prediction parameters for specified prediction indexes are set, and the purpose of accurate prediction is achieved; and finally, the invention solves the problem in the rational use of historical data and determination of the validity of prediction results by using the automatic fitting mean function.

Description

technical field [0001] The invention belongs to the technical field of behavior prediction, and more specifically, relates to a method for predicting a user's behavior in a store based on a Poisson process that automatically fits a mean value function. Background technique [0002] With the development of Internet technology, people's shopping methods have also undergone great changes, and some problems of physical stores have become increasingly prominent due to the development of e-commerce. On the one hand, physical stores often reflect supply and demand information through experience and artificial estimation, which lacks accuracy; on the other hand, there is no reasonable way to objectively reflect customer loyalty, which makes it impossible to capture core customers and provide targeted Serve. Therefore, how to solve the problem of the number of shops in which supply and demand information is unclear, the relationship between supply and demand information of commodity...

AI Technical Summary

Problems solved by technology

Even if time series analysis is only used for users with more frequent purchases, it will face the problem of manual order determination, which is time-consuming and labor-intensive under a huge user base, and the accurate range of prediction results cannot be determined

In addition, the usual forecasting methods do not measure the validity of the forecast results

Moreover, according to the characteristics of a certain user's historical visits to the store, the very important parameter of the Poisson process-the mean function is properly selected and matched, which reduces the reliability and effectiveness of the prediction results

Images