An Improved Local Weighted Mean-Based k-Nearest Neighbor Classifier
Fangfei Liu, Chenqing He, Zhiming Chang · 2023
The article presents a Local Weighted Mean-Based k-Nearest Neighbor (LWMKNN) classifier, which improves the performance of traditional k-nearest neighbor (KNN) and k-nearest neighbor methods based on local means vector. The LWMKNN algorithm incorporates local weighted based on the distance between samples, which enables it to better capture the complex structure of the data. The article provides a detailed analysis of the LWMKNN algorithm, including its underlying principles and mathematical formulations. This provides a clear understanding of how the algorithm works and why it is superior to other KNN methods. The experimental results show that LWMKNN outperforms both KNN and LMKNN on several data sets, demonstrating its effectiveness and versatility. In conclusion, the LWMKNN classifier offers a powerful and flexible tool for machine learning practitioners. By incorporating local weighted, LWMKNN is able to better capture the complex structure of data and outperform traditional KNN methods.