Joint Sparse Estimation Of Coefficient And Covariance Matrix For Multiple Longitudinal Data

We aim to solve the joint sparse estimation of coefficient and covariance matrix for multiple longitudinal data.We proposed to combine CCD algorithm which estimate coefficient matrix and MM algorithm which estimate covariance matrix together and estimate this two matrices at the same time.Previous algorithms are generally about the inverse matrix of covariance and require the nonnegative qualitative.This paper uses the MM algorithm to avoid this constraint and make this algorithm have stronger usability.We use the two types of methods to estimate the regression coefficient matrix.Using simulation studies,we show that algorithm 3 outperforms other algorithm and improve the efficiency of the algorithm.We also use the assessment criteria algorithm RMSE,TPR and TNR to evaluate the performance of the algorithm,which performs just as expected.