IMSTAT Procedure (Analytics)

Example 3: Retrieving Box Plot Values with the NOUTLIERLIMIT= Option

Details

When you specify the NOUTLIERLIMIT= option, the IMSTAT procedure requests outlier information for the variables. When outliers are reported for a variable, pay attention to the last two columns of the display (columns Lo Bin and Hi Bin). These two columns let you know whether the values displayed in the outlier columns are actual data values, or counts in bins. For more information, see the information in the Output section.

Program

libname example sasiola host="grid001.example.com" port=10010 tag='hps';

data example.cars;
    set sashelp.cars;
run;

proc imstat data=example.cars;
    boxplot / noutlierlimit=7; 1

    ods output boxplot=outliers; 2
quit;

proc print data=outliers noobs;
  var column outlo1-outlo5 outhi1-outhi10 binlo binhi;
run;

Program Description

The program example requests that the raw values for up to seven high outliers and up to seven low outliers are retrieved. If there are more than seven outliers, the procedure returns the binned values for the outlying values.
The ODS statement and the PRINT procedure that follows are display purposes only.

Output

In example that follows, using the CARS data set, several variables exhibit outliers on the low end. For example, there are two outlying values for the MPG_City variable. Since the Lo Bin column of the result table is set to No for this variable, the values, 10 for OiutLo1 and 10 for OutLo2, are actual values in the data. Note that these values are smaller than the lower whisker value of 12. (See the previous example.) Similarly, the Horsepower variable shows several outliers on the high end of the distribution and the Hi Bin column is set to No. This lets you know that the values 493, 450, 500, and so on, represent actual values in the CARS table.

On the other hand, the Hi Bin column for the MSRP variable is set to Yes. This lets you know that more outliers were found than the specified NOUTLIERLIMT= limit of 7. The outliers are then placed in bins and the binned counts are reported. For example, there are 14 values in the first bin of MSRP outliers, 8 values in the second bin, 1 value in the fourth bin, and no value in the fifth bin, and so on.

BOXPLOT Statement Results with the NOUTLIERS= Option