International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 12 | Dec 2022
p-ISSN: 2395-0072
www.irjet.net
A comparative analysis of machine learning approaches for movie success prediction Ankit1, Gautam Arora2 1Student, Department of Computer Science And Engineering, SDDIET, Barwala, Haryana
2Assistant Professor, Department of Computer Science And Engineering, SDDIET, Barwala, Haryana
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The success of a movie is crucial for hundreds of people who labour behind the scenes as well as the movie's producers. They rely for their subsistence money generated by the film. The precise foreseeing of a. It's difficult to predict if a movie will be successful or unsuccessful because it has a lot of unknown parameters. In light of this, the machine learning (ML) use in determining if movie will succeed or fail may significantly lower the financial burden shared by all parties. The emphasis of this article is on creating a program that can assist in anticipating the movie's early success will encourage investors to invest analysis is done on some of the patterns from the movie the IMDb collection. Using data gathered from several sources and the system uses a variety of machine learning methods estimates a film's likelihood of success based on its success by looking at historical data from places like IMDb, Reputable Tomato. Experimental findings show that the scores are really outstanding throughout the testing stage. Additionally paper ends by identifying the top actors or actresses in to ensure that the film makes the most money possible. This investigation highlights the value of prediction in the professional realm. Since only these projections serve as the foundation for all capital investments. Key Words—- Movie; Machine Learning; Prediction, Hit; Flop; SVM; k-NN; GNB 1. INTRODUCTION Modern film business is tremendously lucrative, creating a huge area so as to invest. Film investors incur several threats, thus their choice should be extremely carefully considered precise else, they risk incurring enormous debt. Numerous data are accessible from a variety of sources [1]. This planned construction would benefit both the investors and the general public, who may choose whether to view this film or not. The criteria for the success of a film vary depending on the genre. A film's worldwide box office performance, and some Movies may not be as effective at generating income but they can praised for its excellent reviews, ratings, and popularity [2,3]. Various ML methods are used in this paper for predictions. Support Vector Machines (SVM) are them. Both k-Nearest Neighbor (kNN) and Gaussian Naive Bayes (GNB). These algorithms combine the data from actor(s), genre, director, and budget of the film. From the 5000-movie IMDB dataset, movies that have previously produced hits are used to predict future box office success. In this manner, it aids filmmakers in selecting the ideal cast of actors and actresses for any genre. The
© 2023, IRJET
|
Impact Factor value: 8.226
|
outcome of mentioned model is either a hit or a flop. This process verifies each input combination before determining whether to label the film a success or a failure. Movie title, director’s name, actor’s name, and actress name are a few of the characteristics that are entered. There were initially many misconceptions about the traits to choose. We employed feature selection, also known as variable selection, to overcome this problem. In this method, every subset of the variables were selected. It was the most important and essential element. The model has a very high possibility of failing if the wrong qualities were picked. There are 5 parts left in the paper. Part II discusses the pertinent work. The suggested work is shown in part III. The obtained result findings are shown in part IV. The report concludes with a discussion on future research. 2. RELATED WORK There was a lot of study done on this subject in the past. Some earlier efforts used IMDB data to determine their success. Depending on how much money a movie makes, some study divides the work in essentially two categories: hit or failure. We cannot claim that a film's success is only based on its box office performance. The actors, actresses, director, shooting location, screenwriter, music director, etc, all have a role in a movie's success. Some academics calculated the success using historical data. For testing purposes, several studies have made extensive use of NLP systems for collecting movie reviews. Many individuals left reviews for the movie even though they had not seen it on all the screens. Because audience reviews might be skewed by an actor or actress's fan base. In [4], the author created a decision-making system to forecast the box office success utilizing machine learning methods, data mining, and social networks. Their analysis revealed dynamic network connectivity. Their study was mostly based on the elements of who the main actor or actress in the film is, what the film's overall budget is, when it will be released, and how much money the film will ultimately make. They divided the success of movies into three categories: audience, release, and film. Their primary method of forecasting was based on the idea that if the audience is more upbeat, enthusiastic, or happy, the likelihood that the film would be profitable will increase. Similar to this, if a film is more negative and draws fewer
ISO 9001:2008 Certified Journal
| Page 60