Performance evaluation of intelligent prediction models on the popularity of motion pictures

This paper evaluates the performance of intelligent prediction models for predicting the popularity of motion pictures. 113 South Korean movies screened in 2006 are collected and influential input attributes are extracted to build intelligent prediction models, using support vector machines, rough sets and neural networks. 5 different sets of experiments, using 3 additional input attributes, and varying value ranges of output attributes, the number of hidden neurons, the number of training and testing records, and parameter settings of intelligent techniques are conducted to investigate a better accuracy rate of each model. Based on the experimental results, the performance of each model is evaluated and compared with each other to identify a better predictive model on the popularity of movies. The experimental result shows how 5 specific experimental sets affect an accuracy rate of intelligent models for predicting the popularity of motion pictures.