A comparative study of fuzzy linear regression and multiple linear regression in  agricultural studies: a case study of lentil yield management

Ahmad KAZEMIFARD,Shakiba RAJABPOOR,Karim SORKHEH

doi:10.3906/tar-1709-57

Abstract

This study investigates the advantages of two fuzzy linear regression (FLR) models, namely the Tanaka and the Savic and Pedrycz models, over multiple linear regression (MLR) for lentil yield management. We used a fuzzy approach to model the yield of lentil genotypes in which the input is crisp and the output fuzzy. Moreover, after finding FLR equations, we estimated the output corresponding to the collection of fuzzy inputs by using fuzzy algebraic operations and an appropriate defuzzification method known as the center of area method. Results showed the superiority of the Tanaka model over MLR because of reducing the included variables, especially variables available after harvest. The study also emphasizes the advantage of the Savic and Pedrycz model in comparison to the other two models with a smaller error rate.

Highlights

Human beings have been aware of the importance of agricultural products in their life and have benefited from plants and their different parts like fruits, leaves, and seeds, for food, clothing, medicine, and animal feed (Ercisli, 2009; Erturk et al, 2010; Canan et al, 2016; Hricova et al, 2016; Yazici and Sahin, 2016).From among crop plants, lentil (Lens culinaris) from the legume family is an important source of protein, fiber, energy, and minerals for both humans and animals (Torres et al, 2016)
If we look at the relationship between a dependent variable and independent variables from a fuzzy point of view, it is possible to consider some nonlinear relations as “fuzzy linear relations” in classical mathematics
They are as follows: hundred-kernel weight (HKW), pod number (PN), kernel number per pod (KNPP), branch number (BN), leaf area index (LAI), length of the internodes (LI), plant height (H), harvest index (HI), biological yield (BY), days to flowering (DF), total dry matter (TDM), and kernel yield (KY), which were measured at the dry seed stage

Summary

Introduction

Human beings have been aware of the importance of agricultural products in their life and have benefited from plants and their different parts like fruits, leaves, and seeds, for food, clothing, medicine, and animal feed (Ercisli, 2009; Erturk et al, 2010; Canan et al, 2016; Hricova et al, 2016; Yazici and Sahin, 2016).From among crop plants, lentil (Lens culinaris) from the legume family is an important source of protein, fiber, energy, and minerals for both humans and animals (Torres et al, 2016). One of the important areas of research in agriculture is yield management In this area, several investigations have been carried out, such as those of Kravchenko and Bullock (2000), Kitchen et al (2003), Jiang and Thelen (2004), Park et al (2005), and Singh et al (2013). Several investigations have been carried out, such as those of Kravchenko and Bullock (2000), Kitchen et al (2003), Jiang and Thelen (2004), Park et al (2005), and Singh et al (2013) In these studies, different approaches like crop models, statistical tools, and algorithms have been used to evaluate yield. According to Kitchen et al (2003), MLR fails to properly

Objectives

Methods

Results