/* Weed seeds example */
/* Number of weed seeds in 98 quatrer ounce batches 
     of grass. These data are in summary form.
   response variable: number = number of weed 
     seeds in a quarter ounce batche of grass 
   count = frequency   */
/* -------------------------------------------------- */
data weedseeds;
   input number count;
cards;
0 37
1 32
2 16
3 9
4 2
5 0
6 1
7 1
;
/* -------------------------------------------------- */
/* these lines create the SAS data set named "weedseeds"
   the input line names the variables and indicates their 
     order in the data file or datalines 
   the data values are on the lines (datalines) following
   the "cards;" line and the ";" line  
   These data are in summary form with the counts (frequencies)
     for each of the values of the response variable  */          
/* -------------------------------------------------- */
/* The following proc sgplot paragraph requests a 
     histogram for the distribution of 
     number = the number of weed seeds in a batch
   "freq=count" indicates that the variable count contains 
     the frequencies (counts)
   "datalabel=percent" requests that percentages be
      indicates on the histogram
   the "density" command requests a smooth curve to
     approximate the shape of the histogram (a kernel
     density estmate) */
/* -------------------------------------------------- */     
proc sgplot data=weedseeds;
   title 'weed seed distribution';
   histogram number / 
           freq=count datalabel=percent;
   density number / 
             freq=count type=kernel;
/* -------------------------------------------------- */
/* The following proc sgplot paragraph requests a 
     horizontal box plot for the distribution of 
     number = the number of weed seeds in a batch
   "freq=count" indicates that the variable count contains 
     the frequencies (counts) */
/* -------------------------------------------------- */
proc sgplot data=weedseeds;
   title 'weed seed distribution';
   hbox number / freq=count;
/* -------------------------------------------------- */
/* The following proc univariate paragraph requests 
     numerical summary information about the 
     the distribution of number = the number of weed 
     seeds in a batch 
     "var number" indicates that number is the response 
       variable 
     "freq = count" indicates that the data are in summary 
       form with frequencies in the variable count */
/* -------------------------------------------------- */
proc univariate data=weedseeds;
   var number;
   freq count;