The Research Of Resource Access Model Based On Logistic Regression

The traditional resource access model believes that users have “The 80/20 Rule” on the access of video and commodity resources.However,with the development of computer Internet and information technology,the audience is decentralized,video and commodity resources are diversified,and user access patterns have occurred.Change,gradually showing a "long tail effect." How to grasp the law of resource access,and then rational and effective resource allocation and performance optimization for video and commodity resources is a major challenge for video network platform and online shopping platform.In this context,this paper studies the object by the access mode of video and commodity resources,analyzes the early user resource access model,and proposes to use the mathematical statistical regression model Logistic to re-describe the user resource access mode.By analyzing the logistic regression model,users can clearly observe the access rules of video and commodity resources,and can effectively analyze the “hot and cold” degree of video and commodity resources,and then adjust and configure resources to reduce resource scheduling.Loss,so that network resources can be used reasonably and effectively,but also laid the foundation for the study of resource popularity.The main research work and research results of this paper are as follows:(1)Analysed the relevant research information of early video and commodity resource access model(Zipf model),it is found that the current resource access pattern has largely deviated from the early Zipf distribution,and proposed to use Logistic model to re-describe the regularity.(2)Multi-threaded web crawling technology for Youku and YouTube,combined with depth-first strategy and breadth-first strategy for data collection and sorting;and logistic data fitting was performed on the collected video datasets using gradient descent method and least squares method.It is found that the current video access mode can be described using the Logistic regression model.(3)Using the asynchronous crawling framework Scrapy framework,multi-dimensional and multi-cycle data collection for the Jingdong Mall platform,and logistic data fitting was performed on the collected commodity datasets using gradient descent method and least squares method,It was found that the user’s access law to the commodity resources conformed to the Logistic regression model distribution.