arXiv subjects

Iqra Tariq

Publications and source records attributed to Iqra Tariq.

3 recordsLinked to original sources

arXiv2026

Empirical Evidence on Genre-Time Correlation in Box-Office Success Using Exploratory Data Analysis and Machine Learning

The movie industry is one of the fastest-growing global sectors, characterized by high production costs and significant financial risk. Given the capital-intensive nature of filmmaking, accurately predicting box office success is of critical importance for stakeholders ranging from producers to investors. This study investigates the correlation between movie genre and release timing as predictive factors for commercial success. A combined approach involving EDA and supervised machine learning techniques is proposed to assess this relationship. The dataset, comprising the top 200 box office hits and the top 100 flops, was curated from reliable sources, including IMDb, Box Office Mojo, The Numbers, and Wikipedia. EDA revealed that specific genres show statistically significant patterns of success or failure in particular months. For instance, animated and superhero movies achieved their peak success rates in June and July (28% and 29%, respectively), while thrillers and romance genres showed higher hit rates in November. Conversely, the flop dataset showed genres like action and comedy more frequently underperforming in March, April, and August. To validate these findings, multiple regression-based machine learning models were applied using both cross-validation and percentage-split methods. Algorithms such as LWT, Multilayer Perceptron, Random Tree, and Decision Stamp demonstrated high predictive accuracy, reinforcing the hypothesis of genre-time dependency. The results consistently indicated a strong correlation between release month and genre performance, providing valuable insight for strategic planning in content production and release scheduling. This study highlights the growing need to apply data analytics in the media industry, like other data-driven domains, for risk mitigation and optimized decision-making.

cs.CY↗

arXiv2026

The Nexus of Science Fiction, Box Office Success and Technology Representation: A Case Study of the Marvel Cinematic Universe

This paper investigated the applied science domains and subjects depicted in Marvel Cinematic Universe (MCU) movies and assessed the relationship between technological portrayal and box office success. The study looked at 164 publications in academic literature that employed MCU movies. In addition to the foregoing, the study discovered that MCU movies have been used in academic literature in a variety of ways, including teaching science ideas, analyzing ethical dilemmas, and examining social and cultural themes. This shows that MCU movies could be used for educational and societal goals as well. Also, the study demonstrates that MCU movies are more than just popular entertainment. They can also be used to teach science, investigate ethical dilemmas, and investigate social and cultural themes. According to our main investigation, MCU movies represent a wide range of applied science topics, such as technology, magic, ancient technology, cosmic technology, multiverse technology, energy, physics, and engineering. Also, the portrayal of technology is important in the popularity of Science Fiction movies because audiences are drawn to the spectacle of sophisticated technology and the spectacular action sequences that it allows. The study also discovered that the depiction of technology is associated with box office performance, with movies with a higher Tech Content Count being more profitable.

cs.CY↗

arXiv2025

AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection

Deepfake detection has become increasingly important due to the rise of synthetic media, which poses significant risks to digital identity and cyber presence for security and trust. While multiple approaches have improved detection accuracy, challenges remain in achieving consistent performance across diverse datasets and manipulation types. In response, we propose a novel two-tier ensemble framework for deepfake detection based on deep learning that hierarchically combines multiple instances of three state-of-the-art architectures: Xception, Res2Net101, and EfficientNet-B7. Our framework employs a unique approach where each architecture is instantiated three times with different initializations to enhance model diversity, followed by a learnable weighting mechanism that dynamically combines their predictions. Unlike traditional fixed-weight ensembles, our first-tier averages predictions within each architecture family to reduce model variance, while the second tier learns optimal contribution weights through backpropagation, automatically adjusting each architecture's influence based on their detection reliability. Our experiments achieved state-of-the-art intra-dataset performance with AUC scores of 99.22% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.06% (FF++) and 99.94% (CelebDF-v2) without augmentation. With augmentation, we achieve AUC scores of 99.47% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.43% (FF++) and 99.95% (CelebDF-v2). The framework demonstrates robust cross-dataset generalization, achieving AUC scores of 88.20% and 72.52%, and F1 scores of 93.16% and 80.62% in cross-dataset evaluations.

cs.CV↗