Scholarlink

Today’s artificial intelligence (AI), powered by advanced Machine Learning (ML), is achieving remarkable success in various fields. Advanced ML techniques drive it and demonstrate significant achievements across multiple domains. However, these systems’ inherent lack of comprehensibility challenges their implementation in critical sectors. To address this issue, eXplainable Artificial Intelligence (XAI) methodologies have been developed to enhance the interpretability of models for their decision-making processes. This paper presents a comparative analysis of two prominent XAI libraries, SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), along with an evaluation of their performance in conjunction with two Automated Machine Learning (AutoML) tools, Pycaret and TPOT, across five distinct datasets.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Comparative Analysis of SHAP and LIME for Model Interpretability in Automated Machine Learning Using Pycaret and TPOT Across Diverse Datasets

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Parul,
  • Bhawna Jain,
  • Anjali Lathwal

摘要

Today’s artificial intelligence (AI), powered by advanced Machine Learning (ML), is achieving remarkable success in various fields. Advanced ML techniques drive it and demonstrate significant achievements across multiple domains. However, these systems’ inherent lack of comprehensibility challenges their implementation in critical sectors. To address this issue, eXplainable Artificial Intelligence (XAI) methodologies have been developed to enhance the interpretability of models for their decision-making processes. This paper presents a comparative analysis of two prominent XAI libraries, SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), along with an evaluation of their performance in conjunction with two Automated Machine Learning (AutoML) tools, Pycaret and TPOT, across five distinct datasets.