Robust estimating of a quadratic trend's parameter
| Autor: | Alexandra O. Zhukovskaya, Anna V. Kitaeva |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Series (mathematics)
Gaussian Strong consistency Asymptotic distribution Estimator Probability density function 010103 numerical & computational mathematics 01 natural sciences 010305 fluids & plasmas symbols.namesake Robustness (computer science) 0103 physical sciences Convergence (routing) Statistics symbols Applied mathematics 0101 mathematics Mathematics |
| Zdroj: | CoDIT |
| DOI: | 10.1109/codit.2016.7593618 |
| Popis: | We propose new median type estimators of the quadratic trend parameter of location based on the second order differences of observations. Asymptotic normality and strong consistency of the estimators are proved. The estimators have a high breakdown point and high asymptotic efficiency in the Gaussian case, and do not depend on the slope and intercept parameters or their estimates. The simulation results confirm the theoretical ones starting with 20–30 observations. We suppose that the estimators could be used for robust filtering of time series. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|