Nonparametric Spectral Density Estimation for Time Series Data with Missing Values

Missing data;nonparametric spectral density estimation;periodogram;time series;519

자연과학대학 통계학과 ;

University:서울대학교 대학원

关键词: Missing data;    nonparametric spectral density estimation;    periodogram;    time series;    519;   

Others  :  http://s-space.snu.ac.kr/bitstream/10371/131263/1/000000009554.pdf

美国|英语

来源: Seoul National University Open Repository

【 摘 要 】

Spectral density estimation is important in many different areas including astronomy, radio communications, and geophysics. Although many spectral density estimation techniques assume equally spaced complete data, in most cases, real data include missing values. To solve this problem we suggest a new nonparametric spectral density estimation procedure with missing data. Our algorithm is based on the self-consistency and relation between autocovariance and spectral density function, thus it is intuitive and simple. Also, it can be applied with any nonparametric spectral density estimation method such as kernel estimators, wavelet regression, and spline estimators. The practical performance of the proposed method is reported through simulation study.