Machine Learning Models-Based Video Game Sales Prediction

Kaiyue Wu

2024

Abstract

Accurately predicting sales has become particularly important due to the complexity and diversity faced by video game sales forecasting and the impact of multiple factors on sales fluctuations. This paper proposes a novel approach using advanced machine learning algorithms to enhance sales predictions, aiding industry stakeholders in making informed decisions. It employs classical algorithms including linear regression, support vector machines, K-nearest neighbors, random forests, and gradient boosting to analyze global video game sales, based on a dataset of 16,719 games from Kaggle as of December 22, 2020. The methodology involves data preprocessing, feature selection, model evaluation, and hyperparameter tuning, resulting in a streamlined multi-stage prediction process. A new feature selection method is introduced to improve accuracy, while key features are given higher weights to boost performance. Utilizing Gradient Weighted Class Activation Mapping, the results show that the gradient-boosted model surpasses others, delivering precise sales predictions and insights for optimizing inventory, pricing, and marketing strategies. Future research may enhance the model and adapt it for movie or music sales forecasting.

Paper Citation

in Harvard Style

Wu K. (2024). Machine Learning Models-Based Video Game Sales Prediction. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 45-50. DOI: 10.5220/0013205800004568

in Bibtex Style

@conference{ecai24,
author={Kaiyue Wu},
title={Machine Learning Models-Based Video Game Sales Prediction},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={45-50},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013205800004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Machine Learning Models-Based Video Game Sales Prediction
SN - 978-989-758-726-9
AU - Wu K.
PY - 2024
SP - 45
EP - 50
DO - 10.5220/0013205800004568
PB - SciTePress