The Representative Points of Generalized Alpha Skew-t Distribution and Applications

Area of Science:

• Statistics
• Information Theory
• Probability Distributions

Background:

• Statistical distribution is critical in information theory, impacting communication accuracy and entropy.
• Real-world data often deviate from the normal distribution, necessitating flexible models like the generalized alpha skew-t (GAST) distribution to capture skewness.
• Representative points (RPs) are valuable for analyzing complex probability distributions, but their generation can be computationally intensive.

Purpose of the Study:

• To investigate the properties of the GAST distribution, including moment calculations and parameter-peak relationships.
• To explore and compare three methods for generating RPs of the GAST distribution: Monte Carlo (MC), quasi-Monte Carlo (QMC), and mean square error (MSE).
• To evaluate the performance of these RPs in estimating moments and densities for the GAST distribution, considering both known and unknown parameters, and to compare revised and plain maximum likelihood estimation (MLE) methods.

Main Methods:

• Mathematical derivations for GAST distribution properties, including moment calculations.
• Implementation of Monte Carlo (MC), quasi-Monte Carlo (QMC), and mean square error (MSE) methods for generating representative points (RPs).
• Application of RPs for estimating moments and densities of the GAST distribution, and comparison of revised and plain maximum likelihood estimation (MLE) for parameter estimation.

Main Results:

• The study provides proofs for GAST distribution properties, relating parameters to the number of peaks.
• Mean square error (MSE) representative points demonstrated superior performance in estimating moments and densities compared to MC and QMC methods.
• A revised maximum likelihood estimation (MLE) method was found suitable for GAST distributions with unimodal or unobvious bimodal patterns.

Conclusions:

• The GAST distribution is a flexible and adaptable model for various real-world data types.
• MSE-based RPs offer an efficient and effective approach for analyzing GAST distributions.
• The revised MLE method provides a robust parameter estimation technique for specific GAST distribution patterns.