AW K-NN: An Adaptive Weighted K-Nearest Neighbour Framework for Handling Missing Data

Data pre-processing plays a crucial role in ensuring high data quality and reliability during classification.Handling missing values is particularly significant, as the absence of values significantly impacts the performance of imputation methods.The Adaptive Weighted K-Nearest Neighbour (AW K-NN) method is proposed in this research for imputing missing values in datasets.The AW K-NN method utilises distance-based weighting, ensuring that data points closer to each other and potentially more similar have a greater influence, leading to more accurate imputations.After imputing the missing values, a standard scalar is used to standardize the dataset.The classification is performed using three Machine Learning (ML) algorithms namely, Random Forest (RF), Support Vector Machine (SVM) and Gradient Boosting Machine (GBM).These methods are ensembled through soft voting classifier for performance improvement.The datasets used to evaluate the proposed method are the automobile, mammographic and Pima datasets.The AW K-NN imputation method obtained accuracies of 85% on the automobile dataset, 95% on the mammographic dataset, and 82% on the Pima dataset, outperforming existing algorithms like Adaptive Multiple Imputations of missing values utilizing Class Center (AMICC).