Modeling Sparse Data Using MLE with Applications to Microbiome Data

Modeling sparse data such as microbiome and transcriptomics (RNA-seq) is very challenging due to the exceeded number of zeros skewness distribution. Many probabilistic models have been used for modeling data, including Poisson, negative binomial, zero-inflated binomial models. One way identify most appropriate or hurdle based on p-value Kolmogorov-Smirnov (KS) test. The main challenge identifying model that parameters are typically unknown in practice. This paper derives maximum likelihood estimator (MLE) a general class We also derive corresponding Fisher information matrices exploring estimator's asymptotic properties. include new beta Our application shows our more than commonly literature.