Date of Submission
2025-03-14
Supervisor
Dr. Sajjad Haider, Professor, Department of Computer Science, Institute of Business Administration, Karachi
Committee Member 1
Dr. Syed Ali Raza, Assistant Professor, Department of Computer Science, Institute of Business Administration, Karachi
Committee Member 2
Dr. Tahir Syed , Assistant Professor, Department of Computer Science, Institute of Business Administration, Karachi
Degree
Master of Science in Computer Science
Department
Department of Computer Science
School
School of Mathematics and Computer Science (SMCS)
Keywords
SHAP, Feature Selection, LIME, Ensembling models
Abstract
This thesis proposes a novel framework for feature selection aimed at enhancing machine learning model performance, interpretability, and computational efficiency. Feature selection is a critical step in machine learning pipelines, as it directly impacts model accuracy and scalability. However, traditional methods often fail to strike a balance between adaptability across diverse datasets and computational efficiency. This research introduces an integrated approach that combines complementary feature selection techniques to optimize the selection process.
This research presents a robust, model-agnostic, and scalable feature selection method. The proposed framework was evaluated on datasets spanning various domains, including healthcare, finance, and environmental science. Experiments with baseline algorithms such as Random Forest, XGBoost, and Naive Bayes demonstrated that the combined methodology consistently outperforms traditional feature selection methods. The findings show that integrating multiple strategies for feature ranking results in superior performance, particularly when applied to complex, high-dimensional datasets. By focusing on the most impactful features, the framework improves model accuracy and reduces computational overhead without compromising reliability.
Document Type
Restricted Access
Submission Type
Thesis
Recommended Citation
Ahmed, U. (2025). Using explainable AI techniques for feature extraction purposes (Unpublished Unpublished graduate thesis). Retrieved from https://ir.iba.edu.pk/etd-ms-cs/2
