decor: Decorrelate data observations

In XiulinXie/SPCmonitor2: Process Control Chart for Monitoring Short-Range Serially Correlated Data

Description

The function decorrelates a Data Set so that the new data will have no correlation or very small correlations

Usage

decor(x, bmax = 10)

Arguments

x	univariate data
bmax	The smallest number that any data points have no or small correlations with previous data points which means γ(q) ≈ 0 when q > bmax

Details

The input x must be a vector. decor() decorrelate a Data Set based on Cholesky decomposition. It is assumed that the data observations are covariance stationary and the serial correlation exists only when two observations are within bmax > 0 in their observation indices. More specifically, it is assumed that γ(q) = Cov(Xi,Xi+q) only depends on q when i changes, and γ(q) = 0 when q > bmax, where Xi and Xi+q are two process observations obtained at times i and i + q. In practice, the autocorrelation between Xi and Xi+q usually decays when q increases. In such cases, γ(q) is small when q is large, and thus a proper value of bmax can be chosen such that γ(q) ≈ 0 when q > bmax. The default value of bmax is 10, which indicates that the assumption is γ(q) = Cov(Xi,Xi+q) = 0 when q is greater than 10

Value

The data after decorrelation

Author(s)

Xiulin Xie

Examples

decor(rnorm(100,0,1),bmax = 9)

XiulinXie/SPCmonitor2 documentation built on Dec. 10, 2019, 12:10 a.m.