SGPANEL Procedure

LOESS Statement

Creates a fitted loess curve.
About Loess Plots

Syntax

LOESS X= numeric-variable Y= numeric-variable </ option(s)>;

Summary of Optional Arguments

Label options
adds a label for the curve.
specifies the location of the curve label.
displays a label for each data point.
Limit options
creates confidence limits.
specifies the appearance of the mean value confidence limits by using an ODS style element or by specifying fill and line attributes.
specifies the degree of transparency for the confidence limits.
LOESS options
specifies the degree of the local polynomials to use for each local regression.
specifies the degree of the interpolating polynomials that are used for blending local polynomial fits at the kd tree vertices.
specifies the appearance of the fit curve.
specifies the appearance of the markers in the plot.
specifies the maximum number of predicted points for the loess fit and the corresponding limits.
hides the legend entry for the mean value confidence limits.
hides the legend entry for the fit line.
removes the scatter markers from the plot.
specifies the number of iterative reweighting steps to apply to the data.
specifies a smoothing parameter value.
Plot options
specifies the confidence level for the confidence limits.
specifies the value of the ID variable in an attribute map data set.
specifies a classification variable to divide the values into groups.
specifies a label that identifies the fit line in the legend.
specifies a name for the plot.
specifies how observations are weighted.

Required Arguments

X= numeric-variable
specifies the variable for the x axis.
Y= numeric-variable
specifies the variable for the y axis.

Optional Arguments

ALPHA= numeric-value
specifies the confidence level for the confidence limits. Specify a number between 0.00 (100% confidence) and 1.00 (0% confidence).
Default:.05
Interaction:This option has no effect unless you also specify the CLM option.
ATTRID= variable
specifies the value of the ID variable in an attribute map data set. You specify this option only if you are using an attribute map to control visual attributes of the graph. For more information, see Using SG Attribute Maps to Control Visual Attributes.
CLM <= “text-string”>
creates confidence limits. The optional text string overrides the default legend label for the confidence limit.
CLMATTRS = style-element | (CLMAttributeOptions )
specifies the appearance of the mean value confidence limits by using an ODS style element or by specifying fill and line attributes. CLMAttributeOptions can be one or both of the following:
CLMFILLATTRS=style-element | (COLOR= color)
You can specify the color of the fill by using a style element or by using the COLOR= suboption. You can specify colors using the same color schemes that are supported by SAS/GRAPH software. For more information, see Color-Naming Schemes in SAS/GRAPH: Reference.
CLMLINEATTRS= style-element <(options)> | (options)
You can specify the appearance of the outline by using a style element or by using suboptions. If you specify a style element, then you can also specify suboptions to override specific appearance attributes.
The confidence limits have no outlines by default (for most styles). To make the outlines appear, you can change the DISPLAYOPTS in the GRAPHBAND style element. Then you can specify attributes for the outlines. The following code provides an example.
proc template;
  define style MyStyleDefault;
  parent=Styles.Default;
  style GraphBand from GraphBand /
      DisplayOpts = "Fill outline";
  end; 
run;

ods listing style= MyStyleDefault;
proc sgplot data=sashelp.class;
  reg x=height y=weight / CLM
  CLMATTRS=(CLMLINEATTRS= 
   (COLOR=Green PATTERN= ShortDash)); 
run;
The CLMLINEATTRS= options can be one or more of the following:
COLOR= color
specifies the color of the line. You can specify colors using the same color schemes that are supported by SAS/GRAPH software. For more information, see Color-Naming Schemes in SAS/GRAPH: Reference.
Default:For ungrouped data, the default color is specified by the ContrastColor attribute of the GraphConfidence style element in the current style. For grouped data, the default color is specified by the ContrastColor attribute of the GraphData1 ... GraphDatan style elements in the current style.
PATTERN= line-pattern
specifies the line pattern for the line. You can reference SAS patterns by number or by name. For a list of line patterns, see Line Patterns.
Default:For ungrouped data, the default line pattern is specified by the LineStyle attribute of the GraphConfidence style element in the current style. For grouped data, the default line pattern is specified by the LineStyle attribute of the GraphData1 ... GraphDatan style elements in the current style.
THICKNESS= n <units>
specifies the thickness of the line. You can also specify the unit of measure. The default unit is pixels. For a list of measurement units that are supported, see Units of Measurement.
Default:For ungrouped data, the default line thickness is specified by the LineThickness attribute of the GraphConfidence style element in the current style. For grouped data, the default line thickness is specified by the LineThickness attribute of the GraphData1 ... GraphDatan style elements in the current style.
Interaction:This suboption has no effect unless you change the display options in the style element to display outlines. See the preceding code example.
Default:The default appearance of the confidence limits is specified by the GraphConfidence style element in the current style.
Interaction:The CLMATTRS = option has no effect unless you also specify the CLM option.
CLMTRANSPARENCY= numeric-value
specifies the degree of transparency for the confidence limits. Specify a value from 0.0 (completely opaque) to 1.0 (completely transparent).
Default:0.0
Interaction:This option takes effect only if you also specify the CLM option.
CURVELABEL <=“text-string”>
adds a label for the curve. You can also specify the label text. If you do not specify a label, the label from the Y variable is used.
CURVELABELPOS= MIN | MAX | START | END
specifies the location of the curve label. Specify one of the following values:
MIN
places the label at the part of the curve closest to the minimum X axis value.
MAX
places the label at the part of the curve closest to the maximum X axis value.
START
places the curve label at the first point on the curve.
END
places the curve label at the last point on the curve.
Default:END
Interaction:This option has no effect unless the CURVELABEL option is also specified.
DATALABEL <= variable>
displays a label for each data point. If you specify a variable, then the values of that variable are used for the data labels. If you do not specify a variable, then the values of the Y variable are used for the data labels.
Interaction:This option has no effect if you also specify the NOMARKERS option.
DEGREE= 1 | 2
specifies the degree of the local polynomials to use for each local regression. 1 specifies a linear fit and 2 specifies a quadratic fit.
Default:1
GROUP= variable
specifies a classification variable to divide the values into groups. A separate plot is created for each unique value of the classification variable.
Interaction:If you specify the GROUP= option in multiple fit plot statements, then the first GROUP= variable is used for all of the fit plots that specify GROUP=.
INTERPOLATION= CUBIC | LINEAR
specifies the degree of the interpolating polynomials that are used for blending local polynomial fits at the kd tree vertices.
Default:CUBIC
LEGENDLABEL= “text-string
specifies a label that identifies the fit line in the legend. By default, the label “Loess” is used, along with the value of the SMOOTH= option if specified.
LINEATTRS= style-element <(options)> | (options)
specifies the appearance of the fit curve. You can specify the appearance by using a style element or by using suboptions. If you specify a style element, you can also specify suboptions to override specific appearance attributes.
options can be one or more of the following:
COLOR= color
specifies the color of the line. You can specify colors using the same color schemes that are supported by SAS/GRAPH software. For more information, see Color-Naming Schemes in SAS/GRAPH: Reference.
Default:For ungrouped data, the default color is specified by the ContrastColor attribute of the GraphFit style element in the current style.

For grouped data, the default color is specified by the ContrastColor attribute of the GraphData1 ... GraphDatan style elements in the current style.

PATTERN= line-pattern
specifies the line pattern for the line. You can reference SAS patterns by number or by name. For a list of line patterns, see Line Patterns.
Default:For ungrouped data, the default line pattern is specified by the LineStyle attribute of the GraphFit style element in the current style.

For grouped data, the default line pattern is specified by the LineStyle attribute of the GraphData1 ... GraphDatan style elements in the current style.

THICKNESS= n <units>
specifies the thickness of the line. You can also specify the unit of measure. The default unit is pixels. For a list of measurement units that are supported, see Units of Measurement.
Default:For ungrouped data, the default line thickness is specified by the LineThickness attribute of the GraphFit style element in the current style.

For grouped data, the default line thickness is specified by the LineThickness attribute of the GraphData1 ... GraphDatan style elements in the current style.

MARKERATTRS= style-element <(options)> | (options)
specifies the appearance of the markers in the plot. You can specify the appearance by using a style element or by using suboptions. If you specify a style element, you can also specify suboptions to override specific appearance attributes.
options can be one or more of the following:
COLOR= color
specifies the color of the markers. You can specify colors using the same color schemes that are supported by SAS/GRAPH software. For more information, see Color-Naming Schemes in SAS/GRAPH: Reference.
Default:For ungrouped data, the default color is specified by the ContrastColor attribute of the GraphDataDefault style element in the current style.

For grouped data, the default color is specified by the ContrastColor attribute of the GraphData1 ... GraphDatan style elements in the current style.

SIZE= n <units >
specifies the size of the markers. You can also specify the unit of measurement. The default unit is pixels. For a list of measurement units that are supported, see Units of Measurement.
Default:The default marker size is specified by the MarkerSize attribute of the GraphDataDefault style element in the current style.
SYMBOL= symbol-name
specifies the symbol for the markers. For a list of valid marker symbols, see Marker Symbols.
Default:For ungrouped data, the default marker symbol is specified by the MarkerSymbol attribute of the GraphDataDefault style element in the current style.

For grouped data, the default marker symbol is specified by the MarkerSymbol attribute of the GraphData1 ... GraphDatan style elements in the current style.

MAXPOINTS= n
specifies the maximum number of predicted points for the loess fit and the corresponding limits.
Default:201
NAME= “text-string
specifies a name for the plot. You can use the name to refer to this plot in other statements.
Note:The name is case-sensitive. When you refer to the specified name in other statements, you must use the same exact capitalization.
NOLEGCLM
hides the legend entry for the mean value confidence limits.
NOLEGFIT
hides the legend entry for the fit line.
NOMARKERS
removes the scatter markers from the plot.
REWEIGHT= n
specifies the number of iterative reweighting steps to apply to the data.
Default:0
Interaction:This option has no affect if you do not specify the WEIGHT option.
SMOOTH= numeric-value
specifies a smoothing parameter value. If you do not specify this option, a smoothing value is determined automatically.
WEIGHT= numeric-variable
specifies how observations are weighted. Each observation is weighted by a factor of w for computational purposes, where w is the value of the numeric variable. w can be any numeric value. If w is 0, negative or missing, then that observation is excluded from the analysis.

Details

For the SMOOTH= option, the smoothing parameter value must be greater than the minimum value that is determined by the following equation.