Table 1. Comparison matrix of movie box office profitability prediction.

Ref.	Contribution	Technique	Results	Features	Prediction Phase	Dataset
Lash et al. (2015)	Predicted the movie profitability at early stages to increase investor certainty.	Logistic regression	Accuracy 77.1%	Average annual profit, APPG, AWPG, Release dates, Team heterogeneity, Average degree, betweenness centrality, tenure, actor gross	Pre-production	test
Ghiassi, Lio & Moon (2015)	Dynamic Artificial Neural Network (DANN) for the forecasting of movie revenues in the pre-production period.	DANN	Accuracy = 94.1%	MPAA rating, pre-release advertising expenditures, screen count, production budget run time, sequel, seasonality	Post production	test
Zhang et al. (2015)	Uses movie fans as a quantization of director and star value and forecasted movie revenue.	CART	Accuracy (APHR) = 76%	Production Country, Genre, Seasonality, Star value	Pre-production	test
Hunter & Smith (2016)	Examined the movie plot textual properties to predict the movie opening weekend gross.	OLS	MSE = 0.51	Box office revenue, size of the text network	Pre-production	Multiple Sources for Movie Scripts
Lash & Zhao (2016)	Predicted movie profitability to support movie investment decisions at an early phase.	Random Forest	Accuracy = 73%	Basic Features: Genre, Plot Synopsis, Budget, Revenue, Team(Actor, Director)	Pre-production	IMDb and Box Office Mojo
Choudhery & Leung (2017)	Uses tweets and their sentiments to predict box office revenue of the movie.	Polynomial regression model	MSE = 13%	Tweet counts, positive tweets %, negative tweets %, age demographic, sentiment, public reception.	Pre-release, Released	Tweet Dataset
Sachdev et al. (2018)	Effectively estimate the box-office gross revenue for a movie using the public information available after its first weekend of release.	Linear Regression and DTR	MAPE: 24.76%.	Title, genre, release date, budget, No. of screens in OW, OW revenue, IMDB rating, TomatoMeter, TomatoRating, IMDB Popularity, Rotten Tomatoes, and domestic revenue	Released	MDB and Rotten Tomatoes
Mundra et al. (2019)	Performs movie related Tweets sentiment analysis to predict movie success.	Random Forest	Accuracy = 93.17%	Revenue, Budget, Actor 1, 2 and 3, movie Genres, Director, Duration of the movie, No. of users voted, Content rating, Title year, PT/NT ratio	Released	TWeet Dataset
Gao et al. (2019)	Analyze the movie success based on critical and financial perspectives.	SVM	Accuracy = 79.15%	Basic Features: Genre, Plot Synopsis, Budget, Revenue, Team(Actor, Director) and movie Rating	Pre-production	IMDB
Ahmed, Waqas & Afzal (2020)	Forecast box-office success, in the early stage using voting technique	SVM, GdB, XgBoost, RF	Accuracy = 85%	Genre, Director and 3-Actor Rating, Experience,score and FB likes.	Pre-production	Movie Trailer Reviews from Youtube
Kim, Lee & Cheong (2019)	Proposed a deep learning approach using the ELMO embedding and movie sentiment	ELMO, a merged 1D CNN, residual LSTM	F1: 0.68, 0.70	Plot summary text, Genre	Pre-production	CMU Movie Summary Corpus
Ahmad, Bakar & Yaakub (2020)	Using reviews on movie trailers on YouTube, moviegoers intentions to purchase tickets can be predicted.	Multi Linear Regression	Relative AE = 29.65%	Budget, ReviewsCount, ViewsCount, WPNratio, Genre, Rating, LDratio, Duration, PI	Post production	Naver Movies
Zhao, Xiong & Jin (2022)	Movie sales prediction with by evaluating the influence of microblogs.	LR, SVR	Relative AE = 29.65%	Microblogs, Likes, Comments, Forwards	Post production	China Box Office Dataset