Online Program

Clustering Box Office Score Dynamics Using Dynamic Time Warping (303026)

Keywords: dynamic time warping, box office score, time series

The prediction of ticket sales, or box office score, of a movie is a conundrum due to its extreme uncertainty. Many studies have focused on the prediction of final box office scores. Little attention has been paid to the box office score dynamics of ticket sales from opening to closing. The purpose of this presentation is to identify similarities among the dynamics of box office scores using the dynamic time warping (DTW) method. DTW has been widely applied in many applications, including voice recognition, motion detection, species classification, and genetic sequence search. DTW allows us to compare any two time series that may have different time lengths without concerning time series assumptions such as stationarity. To detect clustering patterns of the box office score dynamics, we applied DTW algorithms available in common statistical software tools such as R and SAS. The results will be presented for similarities, along with movie genre, MPAA rating, and other important factors in the film industry.