基于随机森林理论的电影评分预测

Prediction of Film Score Based on Random Forest Theory

电影评分是电影质量的一个直观反映,对未上映的电影评分进行预测是非常重要的。文章在电影本身属性的基础上,定义所有特征量化方式,同时利用电影相似度新增相似电影评分属性,结果表明,加入该因子之后,模型的均方误差降低了35.3%。在此基础上,使用选择性随机森林优化算法对电影评分进行预测,模型的均方误差为0.1025,预测较准确。

Film score is an intuitive reflection of film quality.It is very important to predict the score of unreleased films.Based on the attributes of the film itself,this paper defines all feature quantization methods,at the same time,the film similarity is used to add the scoring attribute of similar films.The results show that after adding this factor,the mean square error of the model is reduced by 35.3%.On this basis,the selective random forest optimization algorithm is used to predict the film score,the mean square error of the model is 0.1025,which is more accurate.