Data Anomalies and Outlier Detection in Chronic Kidney Disease: A Statistical Perspective

To forecast anomalies in the diagnosis of Chronic Kidney Disease (CKD), 400 cases with 25 attributes (11 numerical and 14 nominal) are taken from the UCI Machine Learning Repository. Missing values in numerical and nominal features are addressed by mean and mode imputation techniques respectively. In present work, two methods are utilized to identify outliers: Z-score and the Interquartile Range (IQR). The Z-score method is used to identify extreme departures from the mean; it has shown particularly helpful in identifying outliers in characteristics such as blood pressure and serum creatinine. IQR method, on the other hand identified outliers in traits like age and blood sugar. Using a dual-method approach, this study produces a complete framework for identifying anomalies in data related to Chronic Kidney Disease (CKD). The techniques clarify the existence and severity of CKD by examining the distribution of patient data, which makes early diagnosis and disease progression tracking possible.