Work

Fuzzy Quantile Regression Model With Z‐Numbers and Its Application in an Ambiguous Clinical Situation: Case Study in Heart Disease

Year: 2026

Type: article

Abstract: Nowadays, Z‐numbers are known as the most powerful tool for describing the uncertainty process concerning the degree of confidence. In this matter, regression based on Z‐numbers is called an essential issue because it establishes a functional relationship among Z‐number variables. More importantly, quantile regression is an efficient statistical method for estimating and inferring conditional quantile functions. Thus, this paper aims to develop quantile regression under three models. The first model encompasses explanatory variables, response variable, and Z‐number parameters. The second one includes explanatory variables, a Z‐number response variable, and crisp parameters. Finally, the third one includes crisp explanatory variables, response variable, and Z‐number parameters. The idea of the method for the first model is to employ the mean squared error (MSE) method. On the other hand, the idea of the second model is to utilize a new ranking function. The concept of the third model is to propose a loss function between numbers based on Z‐numbers, which can render some quantiles based on Z‐numbers. Then, a quantile regression model is fitted based on the proposed loss function. Furthermore, two indices are defined to assess the goodness of fit of the optimal quantile regression models based on Z‐numbers. Meanwhile, several examples are solved to assess the efficiency of the proposed models. Eventually, the applicability of the proposed approach is surveyed in heart disease prediction modeling based on a realistically simulated dataset, and future validation on real patient data is recommended. The obtained results reveal that the proposed method outperforms in every condition where regression based on Z‐numbers fails and also provides in‐depth results.

Source: Advances in Fuzzy Systems

DOI: 10.1155/adfs/5195665

Authors: Thaer Hashim Abdulmuttaleb, Reza Hashemi, Somayeh Ezadi

Institution: Razi University

Language: English

Topic: Fuzzy Systems and Optimization

Subfield: Statistics and Probability

Field: Mathematics

Domain: Physical Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $1,000