When you specify CI=SPARSITY or CI=RESAMPLING, the QUANTREG procedure also computes standard errors, values, and -values for regression parameters. Specifies a method to compute confidence intervals for regression parameters. The default is ALPHA=0.05, corresponding to a 0.95 confidence interval. Sets the confidence level for the confidence intervals for regression parameters. See the section Smoothing Algorithm for details. However, the optimal ratio is quite data dependent in practice.
In theory, the smaller the reduction ration, the faster the smoothing algorithm. This ratio is used for reducing the threshold of the smoothing algorithm. RRATIO= value specifies the reduction ratio for the smoothing algorithm. With ALGORITHM=SMOOTH you can specify the following suboption: With the interior point algorithm, you can use the PERFORMANCE statement to enable parallel computing when multiple processors are available in the hardware. See the section Interior Point Algorithm for details. The QUANTREG procedure uses the duality gap as the convergence criterion. TOLERANCE= value specifies the tolerance for the convergence criterion of the interior point algorithm. MAXIT= n sets the maximum number of iterations for the interior point algorithm. However, numeric instability can occur as the parameter approaches 1. The larger the parameter, the faster the algorithm. This parameter should be between 0 and 1. KAPPA= value specifies the step length parameter for the interior point algorithm. With ALGORITHM=INTERIOR you can specify the following suboptions: Table 72.1 Options for Estimation Algorithms
0 kommentar(er)
