Research On E-commerce Intelligent Distortion Based On A/B Test

Under the marketing model of e-commerce,online shopping malls have become a battleground for corporate strategic planning.Relying on the development of big data,web data extraction tools can collect and analyze massive data information such as user clicks and browsing behaviors.A/B testing analyzes user preferences based on this,optimizes websites based on user feedback,and improves corporate interests.At present,the commercial data analysis and integration platforms of domestic enterprises for A/B testing all follow the traditional A/B testing.This paper analyzes the advantages and disadvantages of traditional A/B testing,and also introduces a shell that has not been widely used in commercial use.Compared with the Yeasian A/B test,it is found that the Bayesian A/B test is more suitable for the purpose of enterprise operation,and its commercial value is higher.Further,by classifying the problem of decision-making in A/B testing as the exploration and utilization dilemma in reinforcement learning,and combining the multi-arm gambling machine algorithm with the A/B testing system,a stable and non-linear solution that can be applied to e-commerce business activities is obtained.The intelligent distribution system in a stable environment solves the key problems of intelligent marketing and provides important technical support for building intelligent operations.This paper mainly introduces and compares traditional A/B testing and Bayesian A/B testing,and further combines A/B testing under smooth motion and non-steady motion with the multi-arm gambling machine algorithm to construct a suitable e-commerce promotion activity.intelligent distribution system.Firstly,the specific application of A/B testing in ecommerce enterprises is introduced,and the theoretical and practical applications of A/B testing under the classical frequency theory and Bayesian theory are compared in many aspects.Applicable forms in application environment and different application fields.Secondly,it introduces and analyzes the problems encountered in traditional A/B testing and A/B testing under Bayesian theory in detail,and supports the use of Bayesian A/B testing to improve many of the practical applications of traditional A/B testing.insufficient.Finally,the application of the multi-arm gambling machine algorithm is used to solve the relevant problems that the A/B test algorithm itself is not suitable for practical applications,and for the A/B test data in the environment of smooth and non-stationary action,to construct an intelligent robot based on the multi-armed gambling machine algorithm.Diversion model.The analysis of the test results after modeling shows that the multi-arm gambling machine can improve the A/B test system and achieve intelligent diversion.This research result will greatly reduce the cost of online operation of enterprises,and will also increase sales.The recommended fit of the website to user preferences to further improve the company’s sales performance.