Classification Accuracy of K-Nearest Neighbours Algorithm to Predict Rice Quality

Abstract

This study aims to produce a system that is efficient and effective in predicting the results of a useful decision to determine the accuracy of ricequality with the selection of features/attributes by using Information Gain applied to K-Nearest Neighbor algorithm. One function of the k-nearest neighbor algorithm is as a classification. In its function as a classification in algorithm pattern recognition, k-nearest neighbor is a method of grouping objects based on examples of the nearest training data. This k-nearest neighbor algorithm is the most basic classification technique and is also very simple. In the search for these solutions, several attributes are used for classification. But in its performance there is the possibility of an attribute that does not have a suitable value in data mining, when the attribute is entered into the classification process can cause confusion in the data mining process. Therefore, it is necessary to do attribute selection so that it can identify attributes with values that are irrelevant or relevant and then discard these attributes. In this case, the author uses the information gain method as an identification and also performs attribute selection.