基于线性回归分析法预测李果实干制后果干糖酸比

Prediction of sugar-acid ratio of Prunus salicina L. fruit after drying based on linear regression analysis method

中文关键词:  糖酸比  李果干  多元线性回归

英文关键词:sugar-acid ratio  dried Prunus salicina L. fruit  multiple linear regression

中文摘要:

      目的 基于线性回归分析法建立一种科学地预测鲜李制作成李干后糖酸比的预测模型。方法 以11种李果为实验材料, 利用相关性分析、多元线性回归分析的方法, 探究鲜果的22项具有代表性的理化品质指标与制作成果干后果干糖酸比的关系。结果 以果干糖酸比为因变量Y, 鲜果22项指标为自变量X, 经过逐步筛选挑选出5个重要影响因素: 总糖(X1)、a* (X2)、镁(X3)、可滴定酸(X4)以及可食率(X5), 获得可以用于预测李子果干糖酸比的多元线性回归方程。回归方程决定系数(R2)为0.962, 显著性F检验对应P为0, 有极显著影响(P<0.01)。回归标准化残差分析结果显示, 该方程符合正态分布, 具有较高拟合度。经验证发现, 模型预测值与实际检测值平均相对误差为0.14%, 误差较低。结论 采用多元线性回归模型预测果干糖酸比可行性较高, 预测结果较为精确, 误差较低。

英文摘要:

      Objective To establish a scientific predictive model for the sugar-acid ratio of Prunus salicina L. fruit after being processed into dried Prunus salicina L. fruit based on linear regression analysis method. Methods Eleven kinds of Prunus salicina L. fruits were used as experimental materials. Correlation analysis and multiple linear regression analysis were employed to explore the relationship between 22 physicochemical quality indicators of fresh fruit and the sugar-acid ratio after drying. Results Taking the sugar-acid ratio of dried fruit as the dependent variable (Y) and the 22 indicators of fresh fruit as the independent variables (X), 5 important influencing factors were selected through stepwise screening: Total sugar (X1), a* (X2), magnesium (X3), titratable acidity (X4), and edible rate (X5). Multiple linear regression equations could be obtained to predict the sugar-acid ratio of dried Prunus salicina L. fruit. The coefficient of determination (R2) of the regression equation was 0.962, and the significance F-test corresponding to the P was 0, indicating a highly significant impact (P<0.01). The regression standardized residuals analysis showed that the equation followed a normal distribution and had a high degree of fit. After verification, it was found that the average relative error between the predicted value of the model and the actual detection value was 0.14%, which was relatively low. Conclusion The feasibility of using multiple linear regression models to predict the sugar-acid ratio of dried fruit is relatively high, the prediction results are more accurate, and the error is low.