Abstract: Heterogeneous datasets are prevalent in big-data domains. However, compressing such datasets with a single algorithm results in suboptimal compression ratios. This paper investigates how ...