基于微博数据的电影票房预测模型研究

发布时间：2018-02-04 04:34

本文关键词： 微博数据神经网络线性回归支持向量机电影票房　出处：《安徽大学》2014年硕士论文　论文类型：学位论文

【摘要】：20世纪以来,随着社会的快速发展和人民生活水平的提高,人们追求生活多样化的方式已经不仅限于满足物质生活的要求,还有精神生活的需要。由于影视产业的迅猛发展,电影作为文化产业的重要组成部分,为人们的业务生活提供了乐趣,丰富了人们精神世界。越来越多的人在工作之余不是选择待在家里看电视,而是出来看看电影,感受下新片上映带来的视觉冲击。由于电脑的发明和快速普及,伴随互联网的飞速延伸为影视产业发展提供了强有力的技术支持,3D (Three Dimensions)电影通过配戴立体眼镜,观众可以看到立体的虚拟人物,IMAX (Image Maxium)是一种能够放映比传统胶片更大和更高解像度的电影放映系统。电影的放映种类越来越多,票价也越来越贵,几年间票价由原来的二十元左右涨到100元左右了,投资影视业的利润空间越来越大,如何追求更大的票房收入,成为电影投资者竞相追求的目标。因此,如何提前预测一部电影的票房收入,投资者如何进行前期影视宣传至关重要。电影作为一种生命周期短暂的特殊商品,对其收入的预测难度相对较大,关于电影票房的实证预测一直都很匮乏,也没有一套通用的科学预测方法。如何设计、构建一种科学、可行性强的预测模型,对电影票房进行准确预测,使票房预测真正成为影视行业分析与投资的决定性依据?这些都是当前国内外研究学者们迫切需要解决的难题。本文以新浪微博为研究平台,主要研究了基于微博数据的电影票房预测模型的构建问题。具体工作如下： 1.本文分析了数据挖掘中神经网络、支持向量机和线性回归技术的理论知识,并提出按照数据挖掘标准流程CRISP-DM模型进行票房预测。 2.详细说明了数据提取和准备过程,并且提出分别建立神经网络模型、支持向量机、线性回归模型对票房进行预测,然后进行模型比较,选出最优预测方案的思路。 3.神经网络是功能强大的一般函数预测器。一般而言,神经网络执行预测任务的能力非但不逊于其他技术,有时甚至还有显著优势。还有,它们只需很少的统计或数学知识即可进行训练或应用。实验结果显示,利用神经网络模型在进行票房预测时,平均相对误差很低,预测精度高达90%以上,这体现了神经网络模型预测准确性比其他三种模型高,因此在实际运用中,具有很强的可依赖性,能够为影视票行业提供一个稳定的评估模型。
[Abstract]:Since 20th century, with the rapid development of society and the improvement of people's living standard, people's way of life diversification has been not limited to meet the requirements of material life. Due to the rapid development of the film and television industry, film, as an important part of the cultural industry, provides fun for people's business life. More and more people choose not to stay at home and watch TV after work, but to come out to watch movies and feel the visual impact of the release of new films. With the invention and rapid popularization of the computer, the rapid extension of the Internet has provided strong technical support for the development of the film and television industry. By wearing stereoscopic glasses, viewers can see stereoscopic virtual characters. IMAX Image Maxim is a film screening system capable of showing larger and higher resolution than traditional film. The variety of films is increasing and the price of tickets is increasing. In recent years, the ticket price has risen from 20 yuan to about 100 yuan, the profit space of the investment film and television industry is getting bigger and bigger, how to pursue the bigger box office income, becomes the target which the movie investor competes to pursue. Therefore, how to predict the box office income of a movie in advance, how to carry out the pre-stage film and television publicity is very important. As a special commodity with short life cycle, it is relatively difficult to predict the income of the film. The empirical prediction of film box office has been very scarce, and there is not a set of general scientific prediction methods. How to design, build a scientific, feasible prediction model to predict the film box office accurately. Making box-office prediction really become the decisive basis for film and television industry analysis and investment? These are the current domestic and foreign researchers urgently need to solve the problem. Based on the research platform of Sina Weibo, this paper mainly studies the construction of movie box office prediction model based on Weibo data. The specific work is as follows: 1. This paper analyzes the theoretical knowledge of neural network, support vector machine and linear regression technology in data mining, and puts forward the box-office prediction according to the standard flow of data mining CRISP-DM model. 2. The process of data extraction and preparation is explained in detail, and a neural network model, support vector machine and linear regression model are proposed to predict the box office, and then the model is compared. The idea of selecting the optimal forecasting scheme. 3. Neural networks are powerful general function predictors. Generally speaking, neural networks are not less capable than other techniques, and sometimes even have significant advantages. They require little statistical or mathematical knowledge to train or apply. The experimental results show that the average relative error is very low and the prediction accuracy is more than 90% when using the neural network model to predict the box office, which shows that the prediction accuracy of the neural network model is higher than that of the other three models. Therefore, it has strong dependability and can provide a stable evaluation model for film and television ticket industry.
【学位授予单位】：安徽大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092;TP18

【相似文献】