Research And Application Of Machine Learning In The Analysis Of Film And Television Big Data

As film and television industry become a new breakthrough in the national economy system,the film and television market leading personnel,radio operators,the major video sites operators and a number of research scholars pay more and more attention to it.Faced with the arrival of big data era,the film and television industry faces enormous challenges in certain aspects,such as data storage,processing,analysis and so on.The traditional mode of data storage,data processing method and data analysis techniques will be unable to meet the massive data scenarios.With the continuous development of mathematical and statistical techniques in the field of artificial intelligence,machine learning theory system has been gradually built up.People try to apply machine learning methods to analyze massive data processing,in order to extract useful knowledge and information.Therefore,the study of how to use machine learning methods to dig out the volatility characteristics and trends behind the data from big massive video data is of great practical significance.This thesis is the use of machine learning methods for processing and analysing large film and television data,combined with intelligent television analysis system to carry data preprocessing,feature reduction,chart analysis and ratings prediction,which not only increases the efficiency of data processing,but also improves the accuracy of predictive ratings.Therefore,to solve the problem of the film and television big data scenarios through machine learning methods has important significance,which not only provides the after researchers with effective application ideas,but also makes it possible to win the final market and obtain higher ratings for the film and television industry.In this thesis,the main works are shown as follows:[1]Based on K-Means clustering algorithm for high dimensional film and television data preprocessing,aimed at the screened television sample data,we carry on the attribute selection,data aggregation and data standardization.In which one,we propose using the K-Means algorithm to complete data completion operation.[2]Based on factor analysis of high dimensional film and television data dimension reduction.For complex and high-dimensional feature data,we use factor analysis to obtain low-dimensional and low correlation factors as feature vector after dimensionality reduction.[3]Based on the SVM algorithm and AdaBoost-BP algorithm for classifying television level and predicting television rating.Based on the dimension reduction of the TV series,we use the SVM algorithm and AdaBoost-BP algorithm to train the ratings prediction model,compare and analysis prediction results,sum up the more effectiveness prediction algorithm.[4]Based on intelligent television analysis system to analyze and display.In view of the TV ratings data after processing,we use Echarts to analysis correlation and display intuitively from multi-level and multi-angle.Moreover,the application of the proposed prediction model verify its validity.