delta.star: Compute optimal shrinkage intensities for given thresholding levels
Description
Compute optimal shrinkage intensities for given thresholding levels by minimizing the mean square error.
Usage
delta.star(data, thres.range)
Arguments
data
dataset. A $n$ by $p$ matrix.
thres.range
range of thresholding levels. The default value is thres.range=seq(0,1,by=0.05).
Value
delta.star
the pairs of ($\lambda,\delta*(\lambda))$.
cov
the array of NOVELIST estimators of covariance matrices with the pairs of ($\lambda,\delta*(\lambda))$.
cor
the array of NOVELIST estimators of correlation matrices with the pairs of ($\lambda,\delta*(\lambda))$.
Details
Compute the optimal shrinkage internsity $\delta*(\lambda)$ for any given thresholding level $\lambda$ by minimizing the mean squared error of the NOVELIST covariance estimators to its true covariance matrices.The details of the method are explained by Huang and Fryzlewicz (2015) and Ledoit and Wolf (2003).
References
Huang, N. & Fryzlewicz, P. (2015), "NOVELIST estimator of large correlation and covariance matrices and their inverses". Preprint.
Ledoit, O. & Wolf, M. (2003), "Improved estimation of the covariance matrix of stock returns with an application to portfolio selection." Journal of Empirical Finance 10, 603-621.