Service data processing method and device and related equipment

Abstract

The embodiment of the invention discloses a service data processing method and device and related equipment, and the method comprises the steps: obtaining the target sample user information of a target sample user from a target service sample set, and obtaining a target service object having a binding relationship with the target sample user information; obtaining a prediction service object corresponding to the target sample user information from the initial service prediction model; determining a sample reward business volume according to the target service object, the prediction service object and the maximum prediction service volume, and adjusting an initial service prediction model according to the sample reward service volume, the prediction service object and the target sample userinformation; and when the adjusted initial service prediction model satisfies a target convergence condition, determining the adjusted initial service prediction model as a target service predictionmodel. By adopting the method and the device, the service profit between each user and various service objects can be efficiently and accurately predicted.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a business data processing method, device and related equipment. Background technique [0002] When it is necessary to recommend products to users under the premise of ensuring product revenue, the existing technology establishes a probability prediction model through the mapping relationship between the user information of the user and the status label indicating whether the user's product status has changed. When using the probability prediction model, the user information of the user is input, and the probability prediction model outputs the probability that the user's product state changes. Subsequent probabilities output by the probability prediction model and manual decision-making will recommend corresponding products to users on the premise of ensuring maximum product revenue. [0003] However, the status label of each user can only be obtained by observing the...

AI Technical Summary

Problems solved by technology

[0003] However, the status label of each user can only be obtained by observing the user's product status for a long time, resulting in a small number of user information carrying status labels, which in turn leads to too few sample data carrying labels, and it is impossible to train and predict accurately based on a small number of samples Probabilistic Forecasting Model with Higher Degree

Therefore, the products recommended to users in the future cannot meet the conditions of maximizing product revenue.

At the same time, in the existing product recommendation, it is necessary to manually make a decision to select the recommended product based on the probability of product status changes, resulting in low efficiency in recommending products

Images