발간년도 : [2022]
| 논문정보 |
|
| 논문명(한글) |
[Vol.17, No.6] Algorithm to Handling Incomplete Data from Decision Tree Using SVM |
|
| 논문투고자 |
Jong Chan Lee |
|
| 논문내용 |
This paper examines a method to obtain an estimate that replaces missing information in incomplete data. In other words, the core topic of the proposed paper is to find the estimate most similar to the missing value. To implement this, the training data is classified with the SVM algorithm, and information on the estimate of the missing value is obtained from the decision tree generated in the classification process. Among the various classifiers, the SVM is selected as a classifier for learning because the direction of the classification plane is free and properties having similar distances are gathered at the terminal nodes of the decision tree. From a comprehensive point of view, the training data is divided into lossy data and nonlossy data, and the lossless data is input into the SVM to complete the decision tree. Next, after inputting the loss data into this decision tree, the traversal is repeated until reaching the terminal node according to the condition to find the most similar characteristics. Then, the average value of the lost variable in the terminal node is obtained in three ways, used as an estimate of the loss variable, and the performance of each method is evaluated. All this process is to see if an estimate of the loss value can be derived from some information remaining in the incomplete data. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
