The Research On The Prediction Model Of Box Office Based On FNN

Film box office,as a cultural commodity,is the main indicators of a film's commercial investment success.The significance of the box office prediction is effectively reducing business risk of film investment market,to further promote the industrialization process of the film business management.But the current domestic box office predictions is still in the experience period,this method is too dependent on subjective judgment and expertise,lack of effective data support and theoretical guidance,cannot guarantee the accuracy and stability of the box office predictions,movie parties who mislead rational decisions,while a certain extent,also affected the healthy development of the industry.How to build an accurate and effective box office prediction model for film production marketing investment parties provide a theoretical basis and practical value.The model based on improved fuzzy neural network is proposed in the paper for the problem of this box office prediction which needs fast convergent speed and high accuracy facing large dimension of input data.By utilizing an improved fuzzy neural network algorithm,this model enhances the ability of fuzzy network to deal with fuzzy information and solves the problem caused by non-explanation in neural network.This paper selects 239 films as a data a set.First,on condition that discriminate the influencing factors of the box prediction and considering the feature of film culture and market,the traditional factors affecting the movie box office joined the micro-blog hot word indices,search engines index,press releases and other online data reflect the box office;secondly,a number of factors to predict the box office for the film,this model rapidly and accurately extracts fuzzy rules from complex data by means of the improved fuzzy subtraction clustering algorithm,optimizes the initial parameters of the membership functions and the structure of fuzzy network and make itself more suitable for lightning activity prediction;Finally,according to characteristics of box office prediction,using FNN prediction model based on the Takagi-Sugeno train and test data sets.In the final experiment,the prediction model applies to box office prediction.It's result show that the model is simple,good predictive accuracy,fast speed,to achieve the desired vision paper presented effect.In theory,the model is practical and more important in predicting box office practice.