The judgment method and application of abnormal data in pig performance measurement

In order to improve the accuracy of the performance measurement data and avoid a serious impact of the high proportion of abnormal data on the genetic evaluation results, this study selected breeding data from 2012 to 2018 from some pig breeding farms in Hebei Province as the research object, the structure of the data was analyzed according to the nature and distribution law of the performance measurement data, and a method for judging abnormal data for performance measurement was established. The results showed that abnormal data could be judged by three steps. First, the normal value range was determined according to the normal growth and reproduction law of pigs, and the abnormal data were judged if the range was beyond the normal value.Secondly, if the measurement data was displayed within a reasonable range, the correlation between it and other records should be checked before judging whether it was abnormal data. Finally, the reliability of the overall record was judged. If there were problems with the distribution law of phenotype, genetic analysis was conducted to determine whether the variance components and the estimation results of genetic parameters were within the normal range. The first two types of outliers only account for a small part of the data, which can be used for subsequent analysis after verification or removal. However, if there are problems with the reliability of trait record, it is difficult to judge whether the abnormal data distribution is subjective or objective, so it will not be used in the subsequent analysis.