FyzzyGBR-A gradient boosting regression software with fuzzy target values

Nasiboğlu R., Nasiboğlu E.

SOFTWARE IMPACTS, vol.14, 2022 (ESCI) identifier

  • Publication Type: Article / Article
  • Volume: 14
  • Publication Date: 2022
  • Doi Number: 10.1016/j.simpa.2022.100430
  • Journal Name: SOFTWARE IMPACTS
  • Journal Indexes: Emerging Sources Citation Index (ESCI)
  • Keywords: Gradient boosting, Regression, Fuzzy number, Defuzzification, Fuzzy distance, DECISION TREE, DEFUZZIFICATION, AGGREGATION, DISTANCE, NUMBERS, INFORMATION, OPERATIONS
  • Dokuz Eylül University Affiliated: Yes

Abstract

The Gradient Boosting Regression model is a successful model used in machine learning. In this study, the FuzzyGBR algorithm, which is the novel expansion of the GBR algorithm, in case the target variable contains triangular fuzzy numbers, is discussed and its software developed in Python 3 is explained. Mathematical background required for calculating arithmetic operations and distances on fuzzy numbers is given. The source codes, modules, functions and function parameters of the software are explained. The software was implemented using Python's scikit-learn and seaborn libraries and various datasets from the literature.