A Computational Efficient Fuzzy Clustering Algorithm for Big Incomplete Longitudinal Trial Data

Our previous research has shown that the enhanced Fuzzy C-Means algorithm (eFCM) can improve the clustering quality in the big cross-sectional "synthetic data" [1] where a substantive number of clusters are identified with different degrees of overlap, with enhanced computational efficiency. Here, we provide additional evidence to showcase that our novel initialization method built in MIFuzzy [2-9] [10-12] can improve the computational efficiency in big longitudinal intervention data with missing values. Our numerical analyses further show this improvement in identifying clusters from simulated big incomplete longitudinal data generated using the parameters of our real longitudinal intervention data with missing values. Our findings imply the applicability of this algorithm to similar studies.