.- help for ^adjksm^, ^lorobwei^, and ^roblowes^ (STB-25: snp8) .- Programs for robust scatterplot smoothing ----------------------------------------- ^adjksm^ yvar xvar [^if^ exp] [^in^ range] [^,^ ksm-options ] ^lorobwei^ yvar ysmooth ^roblowes^ yvar xvar bwidth Description ----------- These three programs modify and augment Stata's ^ksm^ command. ^adjksm^ is identical to ^ksm^ except that the bandwidth of the smoother is constant along the x-axis. ^lorobwei^ and ^roblowes^ are used together to perform iterations of Cleveland's lowess smoothing procedure. ^lorobwei^ and ^roblowes^ implement Cleveland's robustness weights, a feature currently not included in Stata's ^ksm^. See help @ksm@ for general information about scatterplot smoothing. Options ------- ^adjksm^ uses the same options as @ksm@. See help @ksm@ for information. ^lorobwei^ and ^roblowes^ do not take any options. Remarks ------- Several steps are required to calculate a lowess smooth using these programs. First, an initial estimate of the smooth is calculated by ^adjksm^. This estimate is used as an input to ^lorobwei^, which calculates robustness weights according to the bisquare weighting function. These weights are stored in a new variable called ^robwei^. As a consequence, you must not have a variable with that name in your data set when you use ^lorobwei^. Then the next iteration of the smoothing process is performed by ^roblowes^, which stores its output in another new variable called ^lowerob^. In schematic form, here is the sequence of steps required: . ^adjksm y x, bwidth(.5) lowess gen(y0)^ /* y0 = initial estimate */ . ^robwei y y0^ /* robwei generated */ . ^lorobwei y x^ /* lowerob generated */ . ^rename robwei w0^ /* save initial weights */ . ^rename lowerob y1^ /* y1 = next estimate */ . ^robwei y y1^ /* begin 2nd iteration */ . ^lorobwei y x^ /* and so on ... */ Authors ------- Isaias H. Salgado-Ugarte Makoto Shimizu University of Tokyo Faculty of Agriculture Department of Fisheries FAX: 011-81-3-3812-0529 Also see -------- STB: STB-25 snp8 Manual: [5s] ksm On-line: help for @ksm@