Histogram per Group of Values Only Greater than Q3

[JonKetchup]

I have a dataset with about 100 groups, and I need a histogram of "Number of Cycles" for each group; however, I only want to display values of "Number of Cycles" that are greater than Q3. Is there a convenient way to do this?

[Reeza]

You can add a WHERE statement/clause to your procedure to filter out the data. How exactly the statement/clause structured depends on your data.
If it's a SAS date for example it could be the following since 4 is the only quarter greater than 3.

where qtr(date) = 4;

[JonKetchup]

Appreciate the help. My variable of interest is not a date; it takes on values from 0 to 1500. Also the Q3 will be different for each group, and I believe a where statement will apply universally to all groups.

[Reeza]

Please provide an example of your data/use case if you need specific coding help. With what you've provided here we can only suggest an approach, as already done.

[JonKetchup]

Here is an example of the data structure. Each group could have as many as 1000 different values of "Number of Cycles", so each group will have a different Q3. Thus, I am looking to create a separate graph for each group, but the histogram will only display values that are greater than Q3 for that particular group. I apologize if my original question was to ambiguous. 

Group	Number of Cycles
Group 1	1
Group 1	9
Group 1	2
Group 1	1
Group 1	10
Group 1	5
Group 2	1
Group 2	2
Group 2	3
Group 2	45
Group 3	2
Group 3	5
Group 3	6
Group 3	4
Group 3	5
Group 3	3

[Quentin · Accepted solution]

Please show an example of your data and the code you have so far.  By Q3, do you mean third quartile / 75th percentile?  If so, then you could calculate the 75th percentile  (by group) using PROC MEANS or whatever and merge it onto your data (by group), then use a where statement Where Var>Var_Q3.  But that's just a wild guess, without seeing sample data.

[Rick_SAS]

Use PROC MEANS to get the values of the Q3 in each group. They merge the Q3 values with the data and output only the values that exceed Q3:

data Have;
input Group	N;
datalines;
1	1
1	9
1	2
1	1
1	10
1	5
2	1
2	2
2	3
2  3
2  4
2  6
2  20
2	45
3	2
3	5
3	6
3	4
3	5
3	3
;

proc means data=Have noprint;
by Group;
var N;
output out=Q3Out Q3=Q3;
run;

data Want;
merge Have Q3Out;
by Group;
if N > Q3;
run;

proc print data=Want;
var Group N Q3;
run;

I don't fully understand if you want a histogram for each group or if you want one histogram that combines all the data across groups.  I think the first. If so, you can use PROC SGPANEL and PANELBY Group, or if you want many small histograms (not paneled), use PROC SGPLOT and BY Group.