Research And Implementation Of Mixed Music Recommendation Algorithm Based On Machine Learning

With the rapid development of the Internet,cloud computing,artificial intelligence and big data technology,technology has profoundly changed people's emotional concepts,value orientations,moral standards,thinking styles and behavior habits.It is extremely convenient for people to obtain information from the Internet,but it also brings the trouble of screening useful information,and the phenomenon of “information overload” has appeared.In order to better overcome and effectively solve this type of problem,on the basis of exploring the existing personalized recommendation algorithms,for the data sparsity and feature extraction problems in traditional recommendation algorithms,this paper proposes a random forest-based extreme gradient lifting The algorithm(RF-XGBoost)'s mixed music recommendation algorithm solves the above problems better,the music recommendation accuracy is higher,and it has better promotion value.The main scientific research work and innovations of this article are as follows:First,in view of the data sparsity problem that has always existed in the recommendation algorithm,dimensionality reduction of the data is one of the main solutions.In this paper,the random forest algorithm improved by the weighted sampling method of attribute subsets is adopted to achieve dimensionality reduction of the data Purpose: When constructing the decision tree in the random forest algorithm,the weighted sampling of attribute subsets can improve the classification strength of the decision tree,and then improve the classification accuracy of the algorithm.Second,for the feature extraction problem in the recommendation algorithm,based on the existing KKBOX music data set,this paper carried out data preprocessing and feature engineering construction on the attributes of the song,the user's operation behavior and song preferences,and portrayed the user portrait,Further improving and enriching user and music information.Third,build a mixed music recommendation algorithm model with random forest algorithm and extreme gradient lifting algorithm as the main body,and use crossvalidation method and grid search strategy to determine the parameters of the algorithm model.The music data set constructed by the algorithm in the second step is predicted.According to the evaluation of the prediction results,it is known that compared with the single model of the random forest algorithm improved by the weighted sampling method of the attribute subset and the extreme gradient lifting algorithm,the The learning performance of the algorithm is better,and the accuracy of prediction is improved by about 3% ? 4%.