Abstract:

Seven main quality indicators of buffalo milk added with different volume ratios of water were measured by
conventional methods based on one-way analysis of variance analysis (ANOVA), principle component analysis (PCA),
and multiple linear regression for quantitative identification of added water in buffalo milk. One-way ANOVA was used to
determine the significant differences in the main quality indicators of different samples with added water to obtain a limit of
detection (LOD) of 7%. According to the PCA, the contribution rate of the first PC was 85.464%, and it included most of the
sample information. There was a significant linear relationship between the scores of the first PC and the content of added
water. Four quantitative model equations were built by multiple linear regression, with correlation coefficients (R2) of 0.965,
0.982, 0.986 and 0.989 and mean absolute errors of −0.23%, −2.40%, 0.23% and 1.28%, respectively, which could be used
for quantitative identification of added water in buffalo milk.

Key words: buffalo milk, added water, physical and chemical indicators, multi statistical analysis, quantitative identification