Rank function missing few ranks

[abhim701]

I am trying to use Rank function on a numeric variable num_var to get a new variable num_var_1 with ranks. In my output I get the new variable num_var_1 with ranks 0,2,3,4,5,6,7,8,9.    Rank 1 is missing.

proc rank data = source_data out = output groups =10 ties = low;
var num_var.;
ranks num_var_1;
run;

What could be the possible reason for this?

Thanks in advance for help.

Regards

Abhishek

[PGStats]

There must be a large number of ties at the minimum value, more than 20% of your data. Ties=low will give rank 0 to those obs.

[abhim701]

Thanks for the help! I got it now.

[abhim701]

I tried ties = high. This also gives me the same output.

0,2,3,4,5,6,7,8,9.

I was expecting 

1,2,3,4,5,6,7,8,9

any comments?

[PGStats]

Can't say without seeing the data. One way to break the ties is to add a small random component to the data:

data source_data_rnd;
set source_data;
num_var_rnd = num_var + rand("UNIFORM")*0.00001;
run;

proc rank data = source_data_rnd out=output(drop=num_var_rnd) groups=10;
var num_var_rnd;
ranks num_var_1;
run;

[Ksharp]

Try
 ties = dense

[Rick_SAS]

 For a discussion of some common reasons this could happen, see the article "Binning data by quantiles."