Application of association rules in nutrition analysis of dairy cow | 关联规则挖掘在奶牛营养研究中的应用

Chinese. 揭示隐含在奶牛营养分析结果数据中影响奶牛瘤胃乙酸/丙酸的关联关系和因素，为奶牛营养研究和奶牛养殖提供参考。选用奶牛营养分析结果的100条数据，应用改进的关联挖掘算法DMApriori算法，对奶牛营养分析结果中瘤胃乙酸/丙酸低于或高于正常值的情况进行了关联分析。饲喂一般牧草黑白花牛的瘤胃乙酸/丙酸低于正常值的可信度为70%；当前体质量在598～698 kg的奶牛多为黑白花牛，可信度为75%；日产奶量在10～20 kg的奶牛多为黑白花牛，可信度为70%；乳脂率在3%～4%的奶牛大多为黑白花牛，可信度为88%。通过对奶牛营养分析中的大量数据进行关联规则挖掘，提取蕴含在这些大量数据中有意义、有价值的信息，可以为奶牛营养研究和养殖的规划调整提供科学依据。

English. This study was done to discover the relationship and factors that affect the value of rumen acetic acid to propionic acid ratio about cow, which may provide-valuable information and scientific ways for producer and technician in field of dairy cow nutrition. One hundred data sets of nutrition analysis result of dairy cow were chosen in this experiment, using the improvement algorithm named DMApriori to analyze the result of nutrition diagnosis of dairy cow and work out what factor may make dairy cow lose balance in value of rumen acetic acid to propionic acid ratio. The results show that the type is China Holstein which is fed on normal grass, it gets lower value than others in rumen acetic acid to propionic acid ratio, the confidence is 70 percent. Cows whose current weights are between 598 and 698 kg are China Holstein, the confidence is 75 percent. China Holstein daily yield milk totally between 10 and 20 kg, the confidence 70 percent, China Holstein yields milk whose fat percentage is between 3 percent and 4 percent, the confidence 88 percent. These results indicate that mining association rules between sets of items in the nutrition analysis of dairy cow database may discover the interesting and valuable information carried in large database, this may help technicians with research on nutrition of dairy cow, assist producer to make layout and modulate for production, and provide scientific basis for them.