NetEase Cloud Music Playlist Forecast Research

In 2015,the country released relevant documents on the issue of music copyright,and the music industry ushered in another opportunity for reshuffle.A large number of users no longer find music through continuous search,but use the song list as the entrance to discover new music.For music platforms,how to use the platform to create high-quality and rich music playlists has gradually become the key to attracting traffic,enhancing the stickiness of platform users and enhancing the value of the platform.Based on the song list information data on the Netease cloud music platform,this paper establishes a statistical model and a machine algorithm model for song list play volume,and predicts the song list play volume.The specific work is summarized as follows:1.Data preprocessing of song list information data.First,it performs data transformation processing on the typed text variables and time variables,and uses word segmentation technology to perform text word segmentation to extract high-frequency words and special vocabularies in text data such as song list names,song list introductions,and labels;Derivative variables of words and special vocabulary.2.Construct a generalized linear model and summarize the characteristics of influencing factors of the playlist volume.This paper firstly analyzes the distribution of data by plotting the histogram,box plot and histogram to reflect the distribution of data,and whether there is a correlation between variables;using normal regression model,gamma regression model,inverse Gaussian regression The model quantitatively searches for deep rules,which summarizes the characteristics of the popular song list,and provides countermeasures and suggestions for the platform and users to create better and rich songs.3.Based on the characteristics of the influencing factors of the playlist summaries,a multiple linear regression,support vector machine,and BP neural network model of the playlist volume is constructed,and the prediction effects of the three methods are compared using cross-validation methods.The results show that from this From the data set,the prediction effect of the support vector machine model is the best,and finally the support vector machine model is applied to the measured data.This article hopes that the analysis can provide Netease cloud music platform with corresponding reference information for predicting the playlist volume,as well as how to create and select high-quality playlists for users of the platform;how to increase the user utilization rate of Net Ease cloud music platform,It has great reference value to enhance user stickiness and push music information to users.