Memprediksi Pendapatan Box Office Film Menggunakan Metode Regresi Deep Learning

Silalahi, Fedinand Putra Gumilang (2025) Memprediksi Pendapatan Box Office Film Menggunakan Metode Regresi Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025201176-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (1MB) | Request a copy

Abstract

Karena kemajuan teknologi dan kemunculan platform digital, prediksi kesuksesan sebuah film menjadi tugas yang semakin penting. Kesuksesan finansial merupakan hal yang penting bagi industri film dan diwakili oleh pendapatan sebuah film. Penelitian ini bertujuan untuk mengembangkan model prediksi pendapatan film dengan menganalisis berbagai faktor yang berpengaruh seperti sutradara, pemeran, kru, kategorisasi genre, dan sebagainya. Memanfaatkan machine learning dan teknik deep learning, penelitian ini menyelidiki keefektivan dari Conditional Generative Adversarial Networks (CGAN), Convolutional Neural Networks (CNN), dan Long Short-Term Memory (LSTM) dalam memprediksi kesuksesan box office film. Penelitian ini memanfaatkan dataset dari IMDb, Box Office Mojo, dan Kaggle, dengan fokus pada film-film berbahasa Inggris yang dirilis secara nasional di Amerika Serikat. Penelitian ini melibatkan pemeriksaan terperinci tentang langkah-langkah preprocessing, rekayasa fitur, dan proses pelatihan model yang diperlukan untuk aplikasi machine learning yang efektif. Penelitian ini juga membandingkan berbagai model untuk menentukan fitur mana yang secara signifikan memengaruhi prediksi pendapatan film dan algoritma mana yang menawarkan prediksi terbaik. Hasil penelitian ini adalah sebuah model yang mampu memprediksi pendapatan sebuah film dengan memasukkan berbagai informasi tentang film tersebut. Dua model ditemukan memberikan hasil terbaik, keduanya memanfaatkan classifier CNN. Model di mana data film disesuaikan dengan inflasi menunjukkan nilai R2 terbaik, sementara model di mana data pelatihan disesuaikan dengan inflasi dan diberi attention layer tambahan menunjukkan error relatif terendah.
===================================================================================================================================
Due to the advancements of technology and the emergence of digital platforms, the prediction of the success of a film has been an increasingly important task. Financial success is important for the film industry and is represented mainly by the revenue of a film. This research aims to develop a predictive model for film revenue by analyzing various influential factors such as directorial credits, cast and crew members, genre categorization, and more. Utilizing machine learning and deep learning techniques, this study investigates the effectiveness of Conditional Generative Adversarial Networks (GAN), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) in predicting the box office success of films. The study leverages datasets from IMDb, Box Office Mojo, and Kaggle, focusing on English-language films released nationwide in the United States. The research involves a detailed examination of the preprocessing steps, feature engineering, and model training processes required for effective machine learning applications. It also compares different models to determine which features significantly impact film revenue predictions and which algorithms offer the best predictive performance. The result of the film is a model capable of predicting the revenue of a film by inputting various information about the film. Two models are found to yield the best result, both making use of the CNN classifier. The model in which the data of the film is adjusted by inflation shows the best R2 score, while the models in which the training data are cleansed from films released before 1980 and films considered to be outliers showcase the lowest relative error.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Box Office, CNN, Conditional GAN, Deep Learning, Machine Learning, Film Revenue Prediction, LSTM
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Ferdinand Putra Gumilang Silalahi
Date Deposited:	30 Jan 2025 01:45
Last Modified:	30 Jan 2025 01:45
URI:	http://repository.its.ac.id/id/eprint/117092

Actions (login required)

View Item