Application of artificial neural network and support vector regression in predicting mass of ber fruits (Ziziphus mauritiana Lamk.) based on fruit axial dimensions

Abstract

Fruit quality attributes are important factors for designing a market for agricultural goods and commodities. Support vector regression (SVR), MLR, and ANN models were established to predict the mass of ber fruits (Ziziphus mauritiana Lamk.) based on the axial dimensions of the fruit from manual measurements of fruit length, minor fruit diameter, and maximum fruit diameter of four ber cultivars. The precision and accuracy of the established models were assessed given their predicted values. The results revealed that using the validation dataset, the developed ANN (R2 = 0.9771; root mean square error [RMSE] = 1.8479 g) and SVR (R2 = 0.9947; RMSE = 1.8814 g) models produced better results when predicting ber fruit mass than those obtained by the MLR model (R2 = 0.4614; RMSE = 11.3742 g). In estimating ber fruit mass, the established SVR and ANN models produced more precise prediction values than those produced by the MLR model; however, the performance differences between the SVR and ANN models were not clear.

Mass modeling of guava (cv. Allahabad safeda) fruit with selected dimensional attributes: Regression analysis approach

“…However, resemblance in external appearance but having the difference in fruit mass makes the grading operation quite challenging; therefore, grading based on mass becomes more significant while designing the advanced machineries. Mass grading of fruits can help to decide the optimum packaging configuration, reduce packaging and transportation costs, and enhance the market potential (Abdel-Sattar, Aboukarima, & Alnahdi, 2021;Seyedabadi, Khojastehpour, Sadrnia, & Saiedirad, 2011). Mass grading can be done by the direct method using a mechanical/electronic weight sizer or by indirect methods.…”

“…Teoh and Syaifudin (2006) developed an algorithm for mango size grading based on the measured area by image analysis. Mass models for apricot (Naderi-Boldaji, Fattahi, Ghasemi-Varnamkhasti, Tabatabaeefar, & Jannatizadeh, 2008), date fruit (Keramat et al, 2008), pomegranate fruit (Mansouri, Khazaei, Hassan Beygi, & Mohtasebi, 2010), cantaloupe (Seyedabadi et al, 2011), persimmon fruit (Shahbazi & Rahmati, 2014), dried ash gourd (Gade, Meghwal, & Prabhakar, 2020), and ber (Abdel-Sattar et al, 2021) based on their dimensional attributes are reported. Mahawar, Bibwe, Jalgaonkar, and Ghodki (2019) established the relevant mass models for graded and ungraded lots of kinnow mandarin based on the physical attributes.…”

Regression models for predicting daily IGS zenith tropospheric delays in West Africa: Implication for GNSS meteorology and positioning applications

“…While in the SRM principle, the generalization accuracy is optimized over the empirical error and the smoothness of the regression function or the capacity of SVM, the ERM principle only minimizes the empirical error and does not consider the capacity of the learning systems, which eventually results in poor generalization performance. Ahmadi and Rodehutscord (2017) and Abdel-Sattar et al (2021) further assert that the SVR model has a wide-ranging approximation capability that can practically approximate all forms of non-linear functions including quadratic functions, and is good at fitting functions and recognizing patterns in diverse kinds of data; whereas the LR models are based on the assumption of linearity, which is useful for only linear approximations, requiring linear function specification to be regressed, hence the flexibility of regression equation may be extremely inadequate. On the other hand, the performance of GAM compared with GLM is also a result of the fact that GAM, as already noted, is an extension of GLM, in which the linear predictor η is not limited to linearity in the covariates X 1 , X 2 , …, X n , but is the sum of smoothing functions s i applied to each covariate X 1 , X 2 , …, X n .…”

Application of image processing and soft computing strategies for non-destructive estimation of plum leaf area

“…SVR is a kernel-based technique in which the kernel function projects the input data into higher-dimensional feature space to find the hyperplane with the lowest error margin and the best fit to the regression line [ 43 , 44 ]. A comparison study conducted by Abdel-Sattar and Aboukarima [ 45 ] proved the superiority of ANN and SVR methods over linear regression methods (LRM) for predicting the mass of Indian jujube fruits based on their axial dimensions. Sabouri and Sajadi [ 46 ] recently reported that using ANFIS and SVR methods, they were able to predict the LA of chia ( Salvia hispanica L.) and quinoa ( Chenopodium quinoa Willd.)…”