<![CDATA[k-Nearest Neighbor (k-NN) is one of the most popular machine learning methods to date. k-NN determines the similarity of data or objects based on the proximity of the distance between data to a class or label or other data group. The distance measurement method on k-NN is considered ineffective because it gives the same weight for each characteristic. The majority vote in determining the data class is also considered problematic because it ignores the closeness between the data and allows multiple majority classes. To solve this problem, the Local Mean Distance Weight k-Nearest Neighbor (LMDWkNN) method emerged. This method is considered capable of providing better accuracy results than k-NN. In this study, we will look at the accuracy results of the LMDWkNN and k-NN methods on the Air Pollution Standard Index (ISPU) data. In this study, 80% of the data was used as training data and the rest was used as testing data. Furthermore, an evaluation will be carried out using the 10-fold cross validation method from the training data so that the best k value will be obtained. The value of k will be used to make predictions on the testing data. The results of this study show that LMDWkNN is able to provide a better accuracy value than k-NN with a more competitive k value. Where the LMDWkNN method found an average accuracy value of 0.9748 while the k-NN was only 0.9371, which means that the LMDWkNN method increased the accuracy value of 0.0377.]]>
Copyrights © 2021
