Movie success prediction

Abstract

The film business is a billion-dollar business, and extensive measure of data identified with motion pictures is accessible over the web. In this system we are analyzing the dataset for predicting the success of the movies. For doing this the analysis of the dataset is done in which the chronicled information of every segment, for example, actor, actress, director, music that impacts the achievement or disappointment of a motion picture is given weight age and after that dependent on different parameters we are predicting whether the movie will be a flop, average or superhit. Certain algorithms are used that can help to predict whether the movies will be a flop, average, or superhit. In this model we focus on the attribute selection for predicting success of the movies. A comparative analysis is to be performed so as to find the accurate results among the algorithms used. Few parameters that are important for predicting success of a movie are gross, genres, release date, star powers of actors, actress, directors, and budget etc. In the dataset there are 28 parameters. The task is to find out most relevant parameters. This will be achieved by Feature selection method as shown in figure 1. Feature selection method is present in “sklearn” library of python. Feature selection method includes Decision trees, information gain, gain ratio. Generating heatmap to visualize success of movie in different regions. Various graphs are generated between time vs algorithms and accuracy vs algorithms for analysis.