Clustering Matrix Variate Data using Parsimonious Mixtures of Skewed Distributions

<title>Abstract</title> <p>Modeling and clustering high-dimensional data through complex mathematical models is often a difficult task and needs a large number of parameters to be estimated. This article presents a family of parsimonious mixture models for matrix variate skewed distributions derived from variance-mean mixtures of matrix variate normal distributions by imposing constraints on the model parameters. This helps in simplifying the complexity and significantly reduces the number of parameters in the model. The Expectation-Conditional Maximization algorithm is employed to estimate the parameters and performed model-based clustering. We evaluate the performance these models using simulated and real datasets. The results demonstrate the validity of parsimonious models in simplifying complex data analysis and also highlight the flexibility in handling smaller sample sizes without compromising clustering accuracy. JEL Classification: C38</p>