ABSTRACT
The research work seeks to find out the efficiency of some methods of estimation of autoregressive model
where the underlying distribution is positively skewed. To examine the performance of estimators on
positively skewed data, data were simulated from various distributions (Weibull, Beta, Gamma, and
Exponential distribution) along different sample sizes. Outliers are one of the major problems affecting
probability distributions and methods of estimations. Outlier points can therefore indicate faulty data,
erroneous procedures, or areas where a certain theory might not be valid. In line with the objectives of this
work, outliers were injected at different percentage (10%, 25% and 50%) in every stage of the simulation
process. Data fitted were estimated using Ordinary Least Square, Maximum Likelihood, BURG and Yulewalker methods of estimation to compare the efficiency of these estimators. Data simulated were from
sample sizes n=5, n=10, n=25, n=50, n=100, n=200, n=500, n=1000, n=2000 and n=5000 were fitted to
check the consistencies of the aforementioned estimators. The shape and scale parameter of Weibull,
Gamma, and Beta distribution were varied at 2 and 1, to check the pattern in which the estimated results
vary. The performance of Beta distribution is better in all the sample sizes irrespective of the orders being
used. Order 3 and 4 under MLE and order 2,3,4 under OLS have the best estimates while the other sample
sizes have small estimates and are the same. Beta distribution performs better than other distributions. The
shape and scale parameter of Beta have little or no effect on the distribution itself.
Keywords: Autoregressive models, Non-Guassian, Outliers, Simulation, Estimation
CITE THIS ARTICLE: Akinyemi SG, Olatayo T, Taiwo IA; Evaluating the Stability of Estimators in NonGaussian Autoregressive Models with Outliers Utilizing Positively Skewed Distribution. Global
Professionals Multidisciplinary Practices Journal. 2024, 1(2): 41-51.
DOWNLOAD FULL DOCUMENT
Akinyemi_Evaluating the Stability of Estimators in Non-Gaussian Autoregressive Models
