A novel region based neighbors searching classification algorithm for big data

• The proposed Region Based Neighbors Searching Classification Algorithm (RNSCA) dynamically segments datasets into regions, enabling precise relationship analysis between test samples and existing points. • Point-wise dynamic searching reduces computational overhead by focusing on region-specific data patterns for efficient exploration. • Integration with ensemble learning enhances classification accuracy by synthesizing diverse regional insights from the dataset. • Rigorous comparisons with state-of-the-art models validate the algorithm’s superiority in handling complex datasets. The K-Nearest Neighbors (KNN) algorithm remains a cornerstone of machine learning due to its intuitive design and effectiveness in classification tasks. However, its performance often suffers from critical limitations, such as sensitivity to the choice of the parameter K and an inability to effectively capture complex relationships among neighboring instances. To overcome these challenges, we propose the Region-Based Neighbors Searching Classification Algorithm (RNSCA) —a novel, adaptive framework that significantly enhances the scalability, flexibility, and accuracy of traditional KNN, especially in high-dimensional and large-scale datasets. RNSCA leverages dynamic, region-based partitioning for more focused and efficient neighbor searches and incorporates a weighted activation function to prioritize the most relevant data points during classification. Additionally, ensemble learning techniques are integrated to strengthen model robustness and improve generalization. The proposed algorithm is extensively validated on benchmark datasets including Iris, Crop Recommendation, Breast Cancer, Diabetes, and Chronic Kidney Disease (CKD). Experimental results consistently demonstrate RNSCA’s superior performance in modeling nuanced local structures and mitigating the core limitations of conventional KNN. This research presents a compelling advancement in classification algorithms, with practical implications across domains such as healthcare, agriculture, and environmental intelligence.