Modeling of non-Gaussian time series in the presence of measurement error

Title

Modeling of non-Gaussian time series in the presence of measurement error

Creator

Thomas, Anna; John, Nimitha

Description

Measurement error in time series refers to inaccuracies or deviations in recorded data that can distort the true underlying patterns, potentially leading to biased estimates and misleading conclusions in statistical analysis. AR models with non-Gaussian errors are crucial for accurately capturing real-world time series data with skewness, heavy tails, or outliers, leading to better forecasts and more reliable statistical inferences. Our focus in this study is on an optimal estimating function (EF) method that accounts for measurement error while estimating the parameters of AR(1) with logistic innovation. The asymptotic properties of the obtained optimal EF are also proved. The simulation study shows that incorporating measurement error into the model yields better estimates compared to ignoring its presence when measurement error exists. This shows that ignoring the measurement error leads to biased estimates. 2026 Taylor & Francis Group, LLC.

Source

Communications in Statistics: Simulation and Computation;

Date

01-01-2026

Publisher

Taylor and Francis Ltd.

Subject

Asymptotic property; Autoregressive; Estimating function; Logistic errors; Measurement error

Coverage

Thomas A., Department of Statistics & Data Science, CHRIST(Deemed to be University), Bengaluru, India; John N., Department of Statistics & Data Science, CHRIST(Deemed to be University), Bengaluru, India

Rights

Restricted Access; Hardcopy may be available in the library

Relation

ISSN: 3610918;

Format

online

Language

English

Type

Article

Identifier

https://doi.org/10.1080/03610918.2026.2625231

https://www.scopus.com/pages/publications/105030248428?origin=resultslist