User selection method and device

Abstract

The embodiment of the invention discloses a user selection method and device. The user selection method comprises: dividing each type of seed user group into a training set and a test set; training seed users in each training set to generate a logistic regression classification model; on the basis of the logistic regression classification model, predicting each seed user included by each testing set to obtain a prediction result; on the basis of the prediction result of each seed user in each testing set and a preset threshold, drawing an ROC curve; acquiring an optimal threshold based on theROC curve; with the logistic regression classification model, predicting each potential user in a potential user group to obtain a prediction result; and selecting a potential user with the predictionresult larger than the optimal threshold from the potential user group as a target user for pushing information. Therefore, the accuracy of the target user selection is improved.

Description

technical field [0001] The embodiments of the present invention relate to the technical field of information processing, and in particular, to a user selection method and device. Background technique [0002] In the prior art, when information delivery (such as product advertisement delivery) is required, the seed user group will be determined first. The so-called seed user group can refer to the user group that has the same demand and interest in products and services, and then the seed user group Diffusion, such as screening out user groups with the same or similar characteristic information (such as user preferences, age, occupation, needs, etc.) Target users. [0003] However, this kind of diffusion without standard means that the target users for information delivery determined by simple screening are not accurate enough. If the selection range of target users is too large, it means that many diffused users will buy the product The probability is very low, which will ...

AI Technical Summary

Problems solved by technology

[0003] However, there is no standard diffusion, and the target users for information delivery determined by simple screening are not accurate enough. If the selection range of target users is too large, it means that many diffused users will buy the product The probability is very low, which will cause a waste of marketing costs, and even cause those users to gradually feel disgusted with the product; on the contrary, if the selection range of target users is too small, it means that many products that should have been spread to Users cannot receive the marketing information, which will lose a large number of potential users and waste business opportunities

Images