DATA SORTING AND LEARNING METHOD, DATA SORTING METHOD, DATA SORTING AND LEARNING DEVICE, DATA SORTING DEVICE, STORAGE MEDIUM STORING DATA SORTING AND LEANING PROGRAM, AND STORAGE MEDIUM STORING DATA SORTING PROGRAM

摘要

PROBLEM TO BE SOLVED: To provide a sorting device suitable for a test data distribution in consideration of the test data distribution as well as a training data distribution and improve sorting accuracy of the test data. SOLUTION: Among test data i, wherein yi is an undefined value and a data NNtrain(i) is correctly learned, a test data iA for providing a minimum d(i, NNtrain(i)) is found out (Step 1). Among test data, wherein yi is an undefined value, a class yNNtest(i) of a data NNtest(i) is not an undefined value, and the data NNtest(i) is correctly leaned, a test data iB for providing a minimum d(i, NNtest(i)) is found out (Step 2). The test data iA and iB are used to be compared with d(iA, NNtrain(iA)) and d(i, B, NNtest(iB)) and the smaller data is made to be iC (Step 3). A label of a most proximate data is substituted for yiC (Step 4). The steps 1 to 4 are repeated for predetermined number of times, and then, the test data is sorted by the data sorting device learned and constituted lastly (Step 5).