International Journal of Computer Applications
Foundation of Computer Science (FCS), NY, USA
|
Volume 187 - Issue 11 |
Published: June 2025 |
Authors: Somapika Das, Radia Iffat Hridy |
![]() |
Somapika Das, Radia Iffat Hridy . Machine Learning-Based Migraine Prediction: Analyzing Key Features and Cause-Effect Relationships for Improved Diagnosis and Management. International Journal of Computer Applications. 187, 11 (June 2025), 13-23. DOI=10.5120/ijca2025925039
@article{ 10.5120/ijca2025925039, author = { Somapika Das,Radia Iffat Hridy }, title = { Machine Learning-Based Migraine Prediction: Analyzing Key Features and Cause-Effect Relationships for Improved Diagnosis and Management }, journal = { International Journal of Computer Applications }, year = { 2025 }, volume = { 187 }, number = { 11 }, pages = { 13-23 }, doi = { 10.5120/ijca2025925039 }, publisher = { Foundation of Computer Science (FCS), NY, USA } }
%0 Journal Article %D 2025 %A Somapika Das %A Radia Iffat Hridy %T Machine Learning-Based Migraine Prediction: Analyzing Key Features and Cause-Effect Relationships for Improved Diagnosis and Management%T %J International Journal of Computer Applications %V 187 %N 11 %P 13-23 %R 10.5120/ijca2025925039 %I Foundation of Computer Science (FCS), NY, USA
Machine learning (ML) has become a critical tool in predictive analytics, enabling accurate and efficient decision-making across various domains. In this study, we evaluated the predictive performance of multiple ML algorithms on a dataset comprising 23 features related to a target outcome. Gradient Boosting Classifier demonstrated the highest accuracy at 90%, followed by Random Forest (87.5%) and Logistic Regression (85%), while Support Vector Classifier (SVC) and k-Nearest Neighbors (KNN) initially showed suboptimal performance, indicating their sensitivity to hyperparameter tuning and feature selection. Feature importance analysis reduced the dataset to 17 significant features, resulting in notable accuracy improvements for SVC (21.25%) and KNN (15%), while Gradient Boosting experienced a slight decline (2. 5%) due to the dependency on excluded features. Logistic regression and random forest remained unaffected, showcasing their robustness across both feature sets. These findings highlight the importance of feature engineering and model optimization in enhancing ML model performance. While Gradient Boosting emerged as a reliable baseline, models like SVC and KNN benefitted significantly from targeted improvements. Future work should focus on expanding the dataset, exploring advanced ensemble approaches, and integrating explainable AI (XAI) techniques to enhance interpretability and reliability. This study provides valuable insights into the role of feature selection and optimization in improving ML-driven predictive analytics for diverse applications.