Scholarlink

<p>The work presented herein focuses on the selection of relevant features for the prediction of severity level of the Chronic Kidney Disease (CKD). The objective is to identify an efficient feature selection algorithm that selects the optimal number of relevant features from a medical dataset to attain better prediction results while diagnosing the severity level of the disease. The statistical insights of the variables and their importance in envisaging the CKD have been studied and the feature “estimated Glomerular Filtration Rate (eGFR)” is extracted to predict the severity of the disease. The ‘class’ of each instance of the enhanced CKD dataset is labelled as per Kidney Disease Improving Global Outcomes (KDIGO) guidelines. Data pre-processing is carried out to handle the outliers effectively by considering the distribution of the data. The enhanced dataset is splitted into training and test datasets in a stratified manner in the ratio of 70:30, and the missing values are handled by considering the type of missingness. Investigation is made on ranking the feature selection algorithms for analyzing the CKD dataset viz., Extra Tree Classifier (ETC), Analysis of Variance, Recursive Feature Elimination, Step Forward Feature Selection, Chi Square, and Mutual Information. Here, Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) with weight optimization is used for ranking feature selection algorithms and the ETC is ranked as top. The ranked feature selection algorithms are validated through classifier models towards the accurate prediction of the severity of the CKD using reduced feature set. The code and resources are publicly accessible at <a href="https://github.com/antonyseba/Enhanced-Feature-Selection-Strategy-.git">https://github.com/antonyseba/Enhanced-Feature-Selection-Strategy-.git</a>.</p>

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

Enhanced feature selection strategy for effective analytics of chronic kidney disease data

  • 期刊论文

  • 发表日期：

    2026年1月6日

查看全文
  • P. Antony Seba,
  • J. V. Bibal Benifa

摘要

The work presented herein focuses on the selection of relevant features for the prediction of severity level of the Chronic Kidney Disease (CKD). The objective is to identify an efficient feature selection algorithm that selects the optimal number of relevant features from a medical dataset to attain better prediction results while diagnosing the severity level of the disease. The statistical insights of the variables and their importance in envisaging the CKD have been studied and the feature “estimated Glomerular Filtration Rate (eGFR)” is extracted to predict the severity of the disease. The ‘class’ of each instance of the enhanced CKD dataset is labelled as per Kidney Disease Improving Global Outcomes (KDIGO) guidelines. Data pre-processing is carried out to handle the outliers effectively by considering the distribution of the data. The enhanced dataset is splitted into training and test datasets in a stratified manner in the ratio of 70:30, and the missing values are handled by considering the type of missingness. Investigation is made on ranking the feature selection algorithms for analyzing the CKD dataset viz., Extra Tree Classifier (ETC), Analysis of Variance, Recursive Feature Elimination, Step Forward Feature Selection, Chi Square, and Mutual Information. Here, Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) with weight optimization is used for ranking feature selection algorithms and the ETC is ranked as top. The ranked feature selection algorithms are validated through classifier models towards the accurate prediction of the severity of the CKD using reduced feature set. The code and resources are publicly accessible at https://github.com/antonyseba/Enhanced-Feature-Selection-Strategy-.git.