Navigation:  Tabs and Options > Miscellaneous Tab >

 

Kernel Density Estimation

 

Print this Topic Previous pageReturn to chapter overviewNext page
Select kernel density estimator for the posterior density: lets you choose the type of kernel density estimator that is used for the marginal posterior density. You can choose between 2 fast and two slow methods. The fast methods are (1) Gaussian, and (2) Silverman-type and Sköld-Roberts correction. The slower methods are (3) Sheather-Jones bandwidth, and (4) Bump killing bandwidth. All these methods are based on a Gaussian kernel. For details, see Sköld and Roberts (2003).
Select kernel density estimator for the prior density: Choice of kernel for the density estimation of the prior density. Possible kernels are: (1) Normal, (2) Epanechnokov, (3) Rectangular, (4) Triangular, (5) Bi-weight, (6) Tri-weight, (7) Laplace, and (8) Logistic. This control is enabled when the check box below has been check marked.
Compute kernel density estimates of prior densities instead of grid density estimates: The prior density can be estimated either by using a grid or by using a kernel density estimator on random draws. The former option is faster and uses the same grid as the check optimum routine; see the Optimization frame on the Options tab.

 

 


Page url: http://www.texlips.net/yada/help/index.html?kernel_density_estimation.htm