Predicting Movie Success based on Machine Learning Using Twitter

Abstract

This paper suggests a method for predicting a box-office success of the film. Lately, as the growth of the film industry, a variety of studies for the prediction of market demand is being performed. The product life cycle of film is relatively short cultural goods.Therefore, in order to produce stable profits, marketing costs before opening as well as the number of screen after opening need a plan. To fulfill this plan, the demand for the product and the calculation of economic profit scale should be preceded. The cases of existing researches, as a variable for predicting, primarily use the factors of competition of the market or the properties of the film.However, the proportion of the potential audiences who purchase the goods is relatively insufficient. Therefore, in this paper, in order to consider people's perception of a movie, Twitter was utilized as one of the survey samples. The existing variables and the information extracted from Twitter are defined as off-line and on-line element, and applied those two elements in machine learning by combining. Through the experiment, the proposed predictive techniques are validated, and the results of the experiment predicted the chance of successful film with about 95% of accuracy.

An Experimental Evaluation of Box office Revenue Prediction through Social Bigdata Analysis and Machine Learning

“…최근 들어 인터넷과 SNS의 발달로 인 해 구전효과를 간접적으로 정량화할 수 있는 방법들이 개발되고 있다 [2][3][4][5] . 이에 따라 대부분의 최신 연구들에서 도 정적 데이터와 동적 데이터를 결합한 예측 기법들이 주를 이루고 있으며, 그 결과로 예측정확도 측면에서도 많은 발전을 가져왔다 [2][3][4][5][6][7][8][9][10] . …”