Imputation and Clustering of Incomplete Data using K-Nearest Neighbors and DBSCAN

Missing values, a persistent challenge in data analysis across various domains, are addressed in this work through a novel approach that combines K-Nearest Neighbors (KNN) imputation and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) for robust clustering. KNN imputation leverages the information from a data point’s knearest neighbors within complete data to estimate missing values, resulting in a more accurate representation of the underlying data distribution. Following imputation, DBSCAN, known for its resilience to outliers and noise commonly encountered in real-world data, is employed for clustering. This combined approach offers a two-fold advantage: 1) effectively handling missing values with KNN imputation, and 2) achieving robust clustering with DBSCAN, particularly beneficial for datasets often containing outliers and missing entries. The effectiveness of this approach is evaluated on multiple datasets (specify the number and types of datasets used), demonstrating its generalizability across different data domains. This study concludes by discussing about the potential future research directions for further exploration.