Analysis of success factors of movies based on Internet Movie Database

Forecast the success of a movie before its release can be very interesting for the film industry but it is difficult to achieve. One technique is to use instance-based algorithms to predict the movie ratings based on a dataset about existing movies. In this thesis, we apply different machine learning methods based on a dataset from the website Internet Movie Database (IMDb) to analyse the success factors of movies in order to predict the movie ratings. The different methods used are the k-nearest neighbors algorithm, the linear regression, the quadratic regression and the k-means clustering. Thanks to these methods and the huge dataset from IMDb, we developed interesting machine learning techniques to foresee the movie ratings with a good accuracy. The techniques used in this thesis could be applied in the film industry to guide the director of a movie to make choices about some features of its movie.