WQD.6 - Classification
WQD.6 - ClassificationSample R code for Classification
nn.result <- knn(WWTrain50S, WWTrain50S, cl=WWTrain5SC$FactQ, k=5)
Nearest neighbor classifier is used with three levels (Low, Medium, High) of quality. It turned out that for k = 5, test data misclassification rate is lowest, when all predictors are being used.
| Quality Classification | |||
| Test Data | Low | Medium | High |
| Low | 293 | 185 | 46 |
| Medium | 267 | 595 | 215 |
| High | 44 | 159 | 233 |
| Accuracy | (293 + 595 + 233) / 2037 = 55% | ||