基于线性回归分析的原蚕种茧产量预测模型建立

Prediction Model of Parent Silkworm Cocoon Weight Based on Linear Regression Analysis

蚕种繁育过程中早期预测原蚕种茧产量有利于科学合理制定一代杂交种的生产计划。利用一元线性回归分析方法及SPSS分析软件,对4个家蚕品种连续20年共181个批次的原蚕熟蚕体质量和种茧产量调查数据进行统计分析,建立以原蚕熟蚕体质量为自变量预测种茧产量的线性回归模型。经F检验分析,该线性回归模型种茧产量预测值与熟蚕体质量调查值之间存在极显著的线性回归关系,且不受养蚕季节的影响,但中系品种与日系品种之间存在极显著差异,故分别建立种茧产量预测模型。利用建立的线性回归模型对2012年春、秋蚕期28个批次的原蚕种茧产量进行预测,预测结果准确,该线性回归模型具有一定的实用性。

During silkworm egg production period, early prediction of the parent silkworm cocoon weight would benefit scientific establishment of production schedule for F1 hybrid eggs. In present study, single -linear regression analysis method and analysis software SPSS were employed to conduct statistical analyses on survey data of mature silkworm larva weight and parent cocoon weight of 4 silkworm varieties which were raised in .181 batches during 20 consecutive years. As a result, linear regression models that use mature parent silkworm larva weight as independent variable to predict parent cocoon weight were established. Statistical Ftests demonstrated that there was extremely significant linear regression relationship between the predicted and observed values. Nevertheless, such relationship was not influenced by silkworm rearing seasons. However, the linear regression models of silkworm varieties of Chinese strain and of Japanese strain varied greatly. Therefore, separate models were established for predicting parent cocoon weight from different strains. Using the established models to predict parent silkworm cocoon weight of 28 batches in spring and autumn rearing seasons of 2012 showed that the prediction was accurate. Thus, the models have certain practicability.