Interval-Valued Data Based Local Mean K-Nearest Neighbors Classifier

The K-Nearest Neighbor (KNN) algorithm is a straightforward and efficient classification technique that exhibits strong performance when addressing fixed data challenges. However, data information is occasionally provided as intervals. To enhance classification performance while dealing with interval-valued data, this paper suggests a novel KNN-based classifier known as the interval-valued data based local mean k-nearest neighbors (ILMKNN) rule. Rather than calculating the direct distances between the variables, the proposed ILMKNN computes the Euclidean distances between the lower and higher bounds of the interval-valued variables. The experiment findings on six UCI data sets show that our suggested classifier performs well in comparison to the interval-valued data-based k-nearest neighbors (IKNN) classifier and is somewhat resistant to diverse selections of k.