body weight prediction of dromedary camels using the machine learning models

The study aimed to compare the accuracy of seven machine learning methods for estimating the weight of dromedary camels, during birth-240 day of age, using the body measurements. with this mind, in overall, 458 records, including body weight and also 12 biometric linear measurements collected from dromedary camels at different stage of life, were used. the seven machine learning methods, including bayesian regu-larized neural network (brnn), extreme learning (el), random forest (rf), support vector machine with linear kernel (lsvm), polynomial kernel (pnlsvm), and radial basis kernel (rnlsvm) and linear re-gression (lr) were compared to estimate the body weight of camels. the performance of the models was evaluated based on mean absolute error, mean absolute percentage error, r-squared, mean squared error, and root mean squared error. a 10 repeated 10-fold cross-validation was used to check the stability of the models and averaged the results. except the tail length, abdomen width, and abdomen to hump height, most predictors had good correlation (r>0.7) with body weight. among predictive variables, the highest correla-tion was 0.96 between heights at whither and height at hump, as well as abdomen width and abdomen to hump height (p<0.01). the accuracy of seven machine learning methods, including brnn, el, rf, lsvm, pnlsvm, rnlsvm and lr were 94.93, 93.22, 94.61, 93.2, 95.43, 94.93 and 93.15, respectively. as final conclusion, the outputs of this report showed that, although all compared models had an acceptable and high performance in predicting the weight based on height of camels, however, the pnlsvm can be suggested candidate model due to expressing the higher accuracy than the others considering all studied criteria.