Iterative Fuzzy C Means, Fuzzy Silhouette, and Imputation for Missing Values in a Dataset

A missing value is an error that always happened, and it is unavoidable. This error should be handled correctly before data is processed into the processing model. This paper proposes a method of imputation by employing iterative Fuzzy C Means (FCM), centroid values and, fuzzy silhouette to handle missing values problem. Missing values can be treated by imputing the missing values. The advantage of FCM is it can provide a better separation of instances when an object is not well separated. It is a well-known clustering method that can provide better clustering result. The optimal clustering value can be measure by using fuzzy silhouette. In this paper, the relationship between imputation based on FCM, fuzzy silhouette and the optimal cluster is identified. Also, the factors that can give impact to accuracy of imputation is recognized, © 2021 IEEE.