A Comparative Analysis of Linear Regression and XGBoost Algorithms for
Predicting GPU Prices Using Technical Specifications

Dublin Core

Title

A Comparative Analysis of Linear Regression and XGBoost Algorithms for
Predicting GPU Prices Using Technical Specifications

Subject

GPU, XGBoost, Linear Regression, Price Prediction, Machine Learning, Technical Specifications

Description

This study investigates and compares the predictive performance of Linear Regression and XGBoost algorithms in estimating Graphics
Processing Unit (GPU) prices based on their technical specifications. GPU prices are known for their high volatility, influenced not only by
hardware characteristics—such as memory capacity, clock speed, and bandwidth—but also by external market factors including demand from
the gaming industry, machine learning applications, and cryptocurrency mining activities. The dataset used in this research comprises 475 GPU
units from three leading manufacturers—NVIDIA, AMD, and Intel Arc—featuring 15 technical attributes obtained from publicly accessible data
sources. Adopting an experimental quantitative approach, the dataset was divided into training and testing subsets using an 80:20 ratio. The data
preprocessing phase involved handling missing values, detecting outliers through the Interquartile Range (IQR) method, performing data
normalization, and encoding categorical features. The models were evaluated using four performance metrics: the Coefficient of Determination
(R²), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The results demonstrate
that XGBoost outperforms Linear Regression, achieving an R² of 0.8129, MAE of 85.07 USD, RMSE of 122.03 USD, and MAPE of 35.23%. In
comparison, the Linear Regression model recorded an R² of 0.7629, MAE of 106.59 USD, RMSE of 137.38 USD, and MAPE of 56.04%. The
superior performance of XGBoost can be attributed to its ability to model non-linear relationships and capture complex feature interactions among
GPU specifications

Creator

Dendi Putra Prakoso1,*
, Muhammad Irfan2
, Quba Siddique3

Source

https://ijiis.org/index.php/IJIIS/article/view/228/152

Publisher

Universitas Amikom Purwokerto

Date

desember 2024

Contributor

Fajar bagus W

Format

PDF

Language

English

Type

Text

Files

Citation

Dendi Putra Prakoso1,* , Muhammad Irfan2 , Quba Siddique3 , “A Comparative Analysis of Linear Regression and XGBoost Algorithms for
Predicting GPU Prices Using Technical Specifications,” Repository Horizon University Indonesia, accessed January 2, 2026, https://repository.horizon.ac.id/items/show/9720.