Abstract: In order to address ambiguity of influence factors in box office prediction, interval number is used to quantify the factors, because interval number is of high utilization of information, scientific and reasonable. In order to ensure objectivity, entropy method is applied to determine weights of each factor. TOPSIS method is used to calculate box office ideal solution nearness degree interval by treating box office influence factors as box office evaluation indicators, and by dividing box office into different categories based on classification of box office. The box office level is determined based on the box office ideal solution nearness degree, and then the interval prediction value of box office is obtained. Combining interval theory, entropy weight method, and TOPSIS method, an interval prediction method for box office before released is obtained. Some action and drama movies released between 2015-2017 are used to verify the proposed method. Average accuracy of the method is 79.33% and 73.92% respectively for action and drama movies, suggesting that the method is effective and practical. The method is valuable for film distributors' decision-making and risk aversion, and has certain reference value for early demand forecast of short life-cycle experience products similar to movies.

Key words: box office prediction before released, interval number, entropy weight method, TOPSIS, interval prediction