FORECASTING FILM FINANCIAL
SUCCESS WITH MACHINE LEARNING
Gokulakrishnan.A Gokulnath.k
Department of IT
Department of IT
Panimalar Institute Of Technology Panimalar Institute Of Technology
Chennai, India
Chennai, India
gokulakrishnantamil@gmail.com gokulnath7707@gmail.com
Miss Rajeshwari,.M.E.,
Assistant professor
Department of IT
Panimalar Institute Of Technology
Chennai, India
Abstract—Predicting society's reaction to
a new product in the sense of popularity
and adaption rate has become an emerging
field of data analysis. The motion picture
industry is a multi-billion-dollar business,
and there is a massive amount of data
related to films that is available over the
internet. This study proposes a decision
support system for film investment sector using
machine learning techniques. This research
helps investors associated with this business
avoid investment risks. The system
predicts an approximate success rate of a
film based on its profitability by analyzing
historical data from different sources like
IMDb, Rotten Tomatoes, Box Office Mojo,
and Metacritic. Using Support Vector
Machine (SVM), Neural Network and
Natural Language Processing, the system
predicts a film box office profit based on
some pre-released features and post-released
features. This paper shows Neural Network
gives an accuracy of 84.1% for pre-released
features and 89.27% for all features, while
SVM has 83.44% and 88.87% accuracy for
pre-released features and all features
respectively, when one away prediction
is considered. Moreover, we figure out that
budget, IMDb votes, and no. Out screens
are the most important features which
play avital role in predicting a film's box
office success.
Keywords— Box office gross;
Data Mining; Machine learning; film
success; film; Critical Predictive analytics;
review; Rating.
I. INTRODUCTION
Predicting society's reaction to a new
product in the sense of popularity and
adoption rate has become an emerging field
of data analysis, and such kind of analysis
can help the film industry to take
appropriate decisions. Can film studios and its
related stakeholders use a forecasting method
for the prediction of revenue that a new
film can generate based on a few given
input attributes like budget, runtime, released
year, popularity, and so on.
1
Ram vilas.H
Department of IT
Panimalar Institute Of Technology
Chennai, India
ramvilas273@gmail.com
This study marks as a decision support system
for the film investment sector using machine
learning techniques. This project helps investors
associated with this business for avoiding
investment risks. The system predicts an
approximate success rate of a film based on its
profitability by analyzing historical data from
different sources like Online rating, Director,
Budget, Pre-Release business, Genre, etc.
The film industry has grown immensely over
the past few decades generating billions of dollars
of revenue for the stakeholders. Now people can
watch movies online and offline on a variety of
mobile devices during leisure or travel through
Netflix, YouTube and downloads. A prediction
system to assess the box office success of new
movies can help the film producers and
directors make informed decisions when making
the film in order to increase the chance of
profitability and box office gross success. New
social media tools are constantly appearing which
are enabling people to gather information on films
and post comments about movies. These
comments can influence the initial prediction
about the box office gross success of a film
which some of the existing research do not
consider. Critic reviews often come out a few
days before the film is released and may,
therefore, help in prediction and at the same time
influence the box office revenue.
II. System Design
EXISTING SYSTEM:
Movies, in general, are products that
have a long development stage until they
reach final consumers and normally at a
high cost level. We can describe a film
development process, in a broader way, as
being composed of 4 (four) stages: Pre-
production, production, post-production and
distribution.The large growth in number of
movies releasing over the past few decades
film Prediction is necessary.
The only way people can check whether the film
will be worth to watch is through applications, so
this system would analyze the reviews posted by
other users, as these reviews are large in number
which the user cannot read and gets confused.
Following are the aims and objectives suggested by
our system. The first step is to identify a dataset of
film data which is suitable for analysis. Relevant
attributes need to be selected from the film data.
Attributes can be general pre production
information regarding film productions such as
film title, sequel, genre, language and information
about writers, actors, and directors. Similarly, the
data must include some measure of success, such as
user film ratings. Secondly, the relevant dataset
has to be prepared and structured in such a way that
the data used is representative of the film scene at
large, as well as suitable for analysis by the
relevant machine learning techniques and
algorithms. Further, correlation is performed on
relevant dataset to find the relationship between al
the variable with each other.
The important step in training our system is to
apply classification model. There are many
classifiers. Lastly, the prediction performance of the
relevant machine learning algorithm has to be
evaluated on the dataset in order to determine
success and failure of film accurately.
III. SYSTEM ARCHITECHTURE
III. MOTIVATION
As data scientists we wanted to dig deeper into the
business side of movies and explore the economics
behind what makes a successful film. Basically we
wanted to
IV. DATA DESCRIPTION
The dataset we utilized to train and test our model
EXPLANATION:
UI (User Interface): This is the space where
interactions between humans and machines
occur. It consists of the hardware and software
that allow effective operation and control of
the machine from the human end, as well as
the exchange of information between the
machine and the human.
Prediction: In the context of machine learning,
prediction refers to the output or result
generated by a model based on the input data.
The model makes predictions by learning
patterns from the training data.

Model: In machine learning, a model is a

PROPOSED SYSTEM: mathematical representation of a real-world
process. It is trained on a dataset to make
This project aims to predict film box office gross predictions or decisions without being
using supervised learning algorithms, specifically explicitly programmed to perform the task.
Hyper parameter Tuning, Decision Tree , Random
Forest. User: This refers to the person or system that
The study found that Random Forest and Hyper interacts with the machine learning model. The
Parameter Tuning had the highest accuracy and can user provides inputs to the model and receives
be used to recommend the best prediction the predictions or outputs.
technique. The project used Jupyter software and
evaluated essential attributes for film gross Inputs: These are the data points or features
prediction. that are fed into the machine learning model.
The model uses these inputs to make
The final results can be displayed using a Flask app predictions.
with an HTML user interface for film box office
gross prediction. Algorithm: This is a set of statistical
processing steps. In machine learning,
algorithms are used to learn from and make
predictions or decisions on data.

Evaluation: This is the process of determining

the performance of a machine learning model.
It involves using various metrics to measure
how well the model is doing in terms of
accuracy, precision, recall, F1 score, etc.

2
 Train Data: This is the dataset that is used to
train the machine learning model. The model
learns patterns from this data to make predictions.

Test Data: This is the dataset that is used to

evaluate the performance of the machine learning
model. The model's predictions are compared
with the actual values in this dataset to measure
its accuracy.

Data Preprocessing: This is the process of

transforming raw data into a understandable
format. It involves cleaning the data, handling
missing values, encoding categorical variables,
scaling numerical variables, etc.

Data CSV: This refers to a file format

(Comma Separated Values) where data is stored
in a tabular format, with each value separated by a
comma. This format is often used for exchanging
data between different applications.
IV. PROPSED METHODOLOGY
A significant part of the dataset required for
the project was extracted from the global TMDB
dataset using its APIs. Following this, the OMDB
API was used to extract the MPAA ratings and
IMDb ratings and votes of each film in the
dataset. The final dataset has the features- Genres,
ID, Original Language, Original Title, Overview,
Popularity Rating, Release Date, Title, TMDB
Rating, TMDB Vote Count, IMDb ID, Budget,
Revenue, Production Companies, Cast, Crew,
Production Countries, Spoken Languages,
Runtime, Tagline, MPAA Rating, IMDb Rating,
IMDb Vote Count and Star Power.
The final dataset has 6065 films. With regards
to Genres, Cast, Crew and Production Companies,
the dataset returned a JSON array of responses.
The popularity index of the most popular cast in
the film was taken as the star power. Only the top
few values in each JSON were considered in the
final dataset since these are the elements along
with star power that attract majority of the
audience.
The extracted dataset was modified, and some
of the es sential features were updated. Features
like ID, IMDb ID, Original title, and Tagline were
not relevant data for the predicting model and
were thus removed. All the NULL values in the
dataset were removed by changing the NULL
values in ’runtime’ to the median value and
replacing the NULL values in the other columns
with an empty set. After removing all the NULL
values from all entries, the ’release date’ feature
was modified by splitting it into three distinct
features for the day, month, and year of the
release. All data except the first three members in
the cast of each entry were deleted.
Then, a check was done for outliers, which are
data points distant from the rest of the data in the
dataset. They have the ability to distort the final
result and prediction, and thus, were removed.
The given dataset was then explored to
understand the relationships between the features
given, how they interact with each other, spot
anomalies in the data, and find patterns to help build
the model. For this purpose, histograms were
plotted to study the range of features like runtime,
budget, popularity, release-data, IMDb rating,
revenue, etc., to study the range of this data.
Following this, a correlation matrix was plotted with
the same features mentioned above to find the linear
interaction between every pair of features. This
contained the correlation coefficient between each
pair. In addition to this, bar plots and frequency
polygons were plotted to study the given data.
V.ALGORITHM AND
TECHNIQUES:
A.Support Vector Machine
Support Vector Machine (SVM) is a supervised
machine learning model that is used for
classification. SVMs work by maximize the
margin between separating hyper plane. In
linear SVM the plane can be split by a line, see
figure 3.14 for an example how the model could
look like. For example, could the red values be
answer A and the blue be answer B. If a new
value would be introduced to the system and
positioned on the red side, the model would
predict the new value to be equal to answer A. If
there are more answers possible a hyper plane is
created to be able to split all the

3
answers up in different areas. SVM are
effective high dimensional, memory efficient,
and versatile machine learning algorithms that
work well with non-linear data.
B.Random Forest
Random Forest or Random decision forests are
an ensemble method for classification, regression
and other tasks that operate by constructing a
multitude of decision trees at training time and
outputting the class that is the mode of the classes
or mean/average prediction of the individual
trees.
VI. EXPERIMENTAL RESULT
The below figures show the results of the module
implementation. These screenshots show the User
Interface through which the modules are being
developed.
Flask:
Flask is a lightweight WSGI web application
framework. It is designed to make getting started
quick and easy, with the ability to scale up to
complex applications.

C.KNN Classification algorithm or K-Nearest

Neighbor algorithm

K-Nearest Neighbor is one of the simplest

Machine Learning algorithms based on
Supervised Learning technique. K-NN algorithm
assumes the similarity between the new case/data
and available cases and put the new case into the
category that is most similar to the available
categories.

D.Linear Regression Home Page:

Linear regression analysis is used to predict the These factors, along with others, can contribute to
value of a variable based on the value of another the overall financial success of a film. However, it's
variable. The variable you want to predict is important to note that the specific combination and
called the dependent variable. The variable you weight of these factors can vary for each film, and
are using to predict the other variable's value is the outcome might not always be predictable.
called the independent variable. This form of
analysis estimates the coefficients of the linear
equation, involving one or more independent
variables that best predict the value of the
dependent variable. Linear regression fits a
straight line or surface that minimizes the
discrepancies between predicted and actual output
values. There are simple linear regression
calculators that use a “least squares” method to
discover the best-fit line for a set of paired data.
You then estimate the value of X (dependent
variable) from Y (independent variable).

Result Page:
You don't need a large time to wait for the results
with in a second the amount will be predicted.

4
CONCLUSION:
The proposed system addresses the limitations
of the existing system by providing accurate
predictions of Film Financial Forecasting using
various supervised learning algorithms. By
comparing the performance of different
classification algorithms, the system identifies the
best-performing technique, ensuring the use of the
most effective method for prediction. The
utilization of Linear Regression, which has been
identified as the algorithm with the highest
accuracy, provides a reliable method for
predicting film box office gross. The user-friendly
interface created using Flask enhances usability
and accessibility, allowing users to input data and
view the predicted box office gross values easily.
FUTURE ENHANCEMENT:
The proposed system can be further improved by
incorporating additional features, such as social
media sentiment analysis, to provide a more
comprehensive prediction of film success. The
system can also be expanded to include real-time
predictions, allowing users to access up-to-date
predictions as soon as they become available.
Additionally, the system can be integrated with
existing film streaming platforms, such as Netflix
or Amazon Prime, to provide personalized
recommendations based on predicted box office
gross.
Furthermore, the system can be scaled up to
analyze larger datasets, allowing for more
accurate predictions and improved performance.
The system can also be adapted to predict the
success of other forms of media, such as music or
books, by using similar supervised learning
algorithms and techniques.
Overall, the proposed system provides a reliable
and accurate method for predicting film box
office gross, and has the potential for further
development and expansion in the future.
REFERENCES
[1] Simonoff, J. S. and Sparrow, I. R. Predicting
film grosses: Winners and losers, blockbusters
and sleepers. In Chance, 2000.
[2] Joshi, M., Das, D., Gimpel, K., and Smith,
N. film Reviews and Revenues: An Experiment in
A.
Text Regression. In Proceedings of the North
American Chapter of the Association for
Computational Linguistics Human Language
Technologies Conference, 2010.
[3] Sharda, R. and Delen, D. Predicting box-office
success of motion pictures with neural networks.
In Expert Systems with Applications, 2006.
[4] “Global box office revenue 2016 |
Statistic.” [Online]. Available: https://
www.statista.com/statistics/259987/global-box-
office revenue/. [Accessed: 03-Jun-2018].
[5] S. Gopinath, P. K. Chintagunta, and S.
Venkataraman, “Blogs, Advertising, and Local-
Market film Box Office Performance,”
Management Science, vol. 59, no. 12, pp. 2635–
2654, 2013.
[6] M. C. A. Mestyán, T. Yasseri, and J. Kertész,
“Early Prediction of film Box Office Success
Based on Wikipedia Activity Big Data,” PLoS
ONE, vol. 8, no. 8, 2013.
[7] J. S. Simonoff and I. R. Sparrow, “Predicting
film Grosses: Winners and Losers,
Blockbusters and Sleepers,” Chance, vol. 13, no.
3, pp. 15–24, 2000.
[8] A. Chen, “Forecasting gross revenues at the
film box office,” Working paper, University of
Washington, Seattle, WA, June, 2002.
[9] M. S. Sawhney and J. Eliashberg, “A
Parsimonious Model for Forecasting Gross Box-
Office Revenues of Motion Pictures,” Marketing
Science, vol. 15, no. 2, pp. 113–131, 1996.
[10] M. T. Lash and K. Zhao, “Early
Predictions of film Success: The
Who, What, and When of Profitability,” Journal
of Management
Information Systems, vol. 33, no. 3, pp. 874–903,
Feb. 2016.
5
[11] A. Sivasantoshreddy, P. Kasat, and A. Jain,
“Box-Office Opening
Prediction of Movies based on Hype Analysis
through Data Mining,”
International Journal of Computer Applications,
vol. 56, no. 1, pp. 1–
5, 2012.

[12] R. Sharda and D. Delen, “Predicting box-

office success of motion
pictures with neural networks,” Expert Systems
with Applications,
vol. 30, no. 2, pp. 243–254, 2006.