基于背膘厚度构建民猪肌内脂肪含量的预测模型  

Construction of prediction model of intramuscular fat content based on backfat thickness of Min pigs

作　　者：刘亚欣 何鑫淼[2] 刘欣睿 秦婕 王嘉博 王文涛[2] 吴赛辉[2] 刘娣 钟金城[1] LIU Yaxin;HE Xinmiao;LIU Xinrui;QIN Jie;WANG Jiabo;WANG Wentao;WU Saihui;LIU Di;ZHONG Jincheng(Key Laboratory of Qinghai-Tibetan Plateau Animal Genetic Resource Reservation and Utilization,Sichuan Province and Ministry of Education,Southwest Minzu University,Chengdu 610041,China;Institute of Animal Husbandry,Heilongjiang Academy of Agricultural Sciences,Harbin 150086,China)

机构地区：[1]西南民族大学、青藏高原动物遗传资源保护与利用四川省教育部重点实验室,成都610041 [2]黑龙江省农业科学院畜牧研究所,哈尔滨150086

出　　处：《黑龙江畜牧兽医》2023年第13期63-67,135,共6页Heilongjiang Animal Science And veterinary Medicine

基　　金：中央高校基本科研业务专项基金项目(2022NYXXS047);四川省科技支撑计划项目(2021YJ0269,2021YJ0266);国家肉牛牦牛产业技术体系项目(CARS-37);青海省科技计划项目(2021-ZJ-736);黑龙江省科研业务费项目(CZKYF2020A004,CZKYF2021D003);黑龙江省农业科学院院杰青科研项目(2021YYYF021)。

摘　　要：为了探究不同统计模型预测猪肌内脂肪含量和眼肌面积的可行性,试验利用测定的312头民猪的体重、背膘厚和日龄数据,以岭回归最佳线性无偏估计(ridge regression best linear unbiased prediction,rrBLUP)、贝叶斯B (Bayes B)、随机森林算法(Random Forest,RF)三种模型来预测民猪肌内脂肪(intramuscular fat,IMF)含量和眼肌面积(eye muscle area,EMA),每种模型均采用5倍交叉法和去一法进行验证,比较预测结果以得出预测准确率较好的一种模型。结果表明:两种验证方法中,去一法的预测准确率要明显高于5倍交叉法,但计算效率较慢。在预测肌内脂肪含量的模型中,rrBLUP、Bayes B、RF模型的预测准确率分别为0.639,0.595,0.631,其中rrBLUP模型预测效果较好。在预测眼肌面积的模型中,rrBLUP、Bayes B、RF模型的预测准确率分别为0.618,0.464,0.642,其中RF模型预测效果较好。说明基于体重、日龄和背膘厚测定数据预测民猪IMF和EMA是可行的。In order to explore the feasibility of intramuscular fat content and eye muscle area in pigs predicted by different statistical models,based on the body weight,backfat thickness and age data of 312 Min pigs,the intramuscular fat(IMF) content and eye muscle area(EMA) of Min pigs were predicted by Ridge regression best linear unbiased prediction(rrBLUP),Bayes B and Random Forest(RF).Each method was verified by the 5-fold crossover method and the leave-one-out method,and the method with better prediction accuracy was obtained by comparing the prediction results.The results showed that the prediction accuracy of the leave-one-out method was significantly higher than that of the 5-folds cross verification in the two verification methods,but the computational efficiency was slower.In the model for prediction of intramuscular fat content,the prediction accuracy of rrBLUP,Bayes B and RF models were 0.639,0.595 and 0.631 respectively.The rrBLUP model had a better prediction effect.In the model for prediction of eye muscle area,the prediction accuracy of rrBLUP,Bayes B and RF models were 0.618,0.464 and 0.642,respectively.The prediction effect of RF model was better.These results indicated that the projected IMF and EMA of Min pigs based on the measurement data of body weight,days of age and backfat thickness were feasible.

关 键 词：民猪 构建模型 肌内脂肪 眼肌面积 预测准确率 岭回归最佳线性无偏估计模型 贝叶斯B模型 随机森林算法模型 

分 类 号：S828.89[农业科学—畜牧学]