| With the improving of people's living standard and the rapid development of the economics,"Food Safety" becomes a hot topic,which is also the main concern of government.Lager amounts of data have been produced with the spot check on food quality and safety,which is a significant method to ensure food quality.How to make the data contribute to spot check on food quality and safety become a problem.Hence,to provide suggestions of selecting key samples to guide the spot check for relevant departments as well as to reduce expenditure of labor,material and financial,it is important to take appropriate algorithms to analyze the data to figure out the potential relevancy between food attribute and food quality.Based on food sampling data supported by National Food and Drug Administration.In this study,after preprocessing the dataset,two types of tagged sample which marked "qualified samples" and"unqualified samples" were selected.In addition,we treated the food category,food production season,food production place,food sampling season,food sales place as the training set of classification algorithm.By analyzing and comparing 4 common machine learning classification algorithm,for instance,Logistic Regression Algorithm,C4.5 Decision Tree Algorithm,BP(Back Propagation)Neural Network Algorithm and Support Vector Machine Algorithm.Finally,in accordance with the experimental results,the Support Vector Machine algorithm is selected as the main algorithm to predict the quality of food sampling.In this paper,the after-training prediction model is applied in the function of food quality prediction,together with the national food sampling data analysis,which aimed to establish national food sampling data analysis and mining system.The system has the characteristics of simple operation,strong function and friendly interface.To some extent,it can assist the relevant departments of food sampling to complete the work of food sampling and provide decision-making opinions. |
PDF Full Text Request