Research On The Optimization Of Materials Classification Based On Cost-Sensitive Learning

Materials classification is the material was classified into different categories of product or different products in accordance with the requirements of the fixed standard or a certain requirements to increase its use value to meet the needs of the consumption, production or bring so that convenience to the value-added process of current trading. Material classification is to induce the non-sorted material into corresponding target class based on the characteristics of materials classification. With the perfection of test technology and data classification of all kinds of material classification, the cost or cost optimization in the process of materials classification more and more becomes an important part of the study of material classification.Paper first reviews the testing technology and related algorithms of material classification characteristics, the analysis of the economic benefits of material classification showed that the test cost and misclassification cost in the process of material classification has a direct and significant effect on the economic benefits of material classification, on the basis of which the mathematical model of material classification optimization problem is established, and points out material classification optimization problem is to consider test cost and misclassification cost, to further reduce the cost of material classification process, and improve the economic benefits of material classification on the basis of achieving a certain material classification accuracy.Synthesizing the advantages and disadvantages of ID3, C4.5, EG2and C4.5CS algorithms, this paper proposed EC4.5CS cost-sensitive learning algorithm, and designed the solving algorithm of enhanced information cost function and the related parameters of EC4.5CS algorithm in detail. In the end, making apple classification as an example, the results of the data experiments and verification of the material classification optimization algorithm based on EC4.5CS showed that the total cost of materials classification algorithm based on EC4.5CS is much better than that of the ordinary classification algorithm (ID3) not considering the cost, and further improve the apple classification accuracy and reuce the classification cost through resampling technology finally. The final experimental results show that the reduction rate of the total cost based on the EC4.5CS algorithm and double sampling technology is as high as44.93%compared with ID3algorithm. Material classification is the key content of product quality standards and a core link of postpartum processing and commercialization shifting as well. At present, the implementation scope of material classification in our country is not large, and the degree is generally low, the main reason is that the cost of material classification is larger than the economic benefits of material classification. The research in this paper provides a research idea for the cost optimization of material classification process, which has a positive role in expanding the application range of the material classification in our country.