Baby formula classification based on forth order polynomial smoothing support vector machine
Yu-Bo Yuan, Jing Lu, Feilong Cao · 2011
The quality of baby formula is of great importance for babies to get adequate and balanced nutrition, and grow healthily. But how to tell the good from the bad is always a problem which perplex the parents. So in this paper, we try to use a mathematical method, the support vector machine method, and from the angle of nutritional components and their contents, to find a way to solve this problem. The reason why to choose polynomial smooth support vector machine model is that compared with the traditional one, it can have better performance in classification. In this paper, we first use the evaluations of baby formula which are acquired through the Internet to train the support vector machine, but the results are not good. Then we use the cluster method, to put the baby formula into two kinds, combine with the evaluation, training again, at this time the performance improved much. According to the above method, we classify the baby formula whose classifications are unknown, and the conclusions are as follow: some kinds of baby formula which are not thought of high quality are actually in similar components and contents with those international brands. Several domestic brands though not very famous, but behind them, we can see good quality and consumer's trust for them.