topic Does "definekey" enforce uniqueness in a hash object? in New SAS User

Does "definekey" enforce uniqueness in a hash object?

Syntas_error — Mon, 20 May 2019 12:25:02 GMT

From a practice exam:

-----------

The dataset Cars contains observations listing hte Make, Model and Country of Origin of 300+ vehicles. Consider the SAS code below:

data_null_;

attrib Make length=$15 Label="Make Name";

declare hash Makelist(dataest:'cars', ordered:'yes');

Makelist.definekey('Make');

Makelist.definedone();

rc=Makelist.output(dataset: 'Manufacturer');

run;

What is the result of running this program?

A) The manufacturer dataset contains unique values of Make in ascending order

B) The manufacturer Dataset contains unique values of Make in the original order as the input data.

C) The manufacturer Dataset contains a copy of the Cars dataset in order

D) The manufacturer Dataset contains a single observation as the data step iterates only once.

I answered C but the correct answer is A. Does this then mean that the definekey statement enforces uniqueness on those variable values, similarly to the PRIMARY KEY constraint?

Re: Does "definekey" enforce uniqueness in a hash object?

PeterClemmensen — Mon, 20 May 2019 13:31:22 GMT

Yes, the correct answer is A.

First of all, only the Make variable is read into the hash object. Therefore, the manufacturer data set can not contain a copy of the Cars data set.

Secondly, when the Duplicate Argument Tag is not specified with the value 'Y', only the first value for each unique value of Make is read into the hash object. If you specified the code like below, all the values of Make (including duplicates) will be read into the hash object and consequently a part of the manufacturer data set

data _null_;

attrib Make length=$15 Label="Make Name";

declare hash Makelist(dataset:'sashelp.cars', multidata:'Y', ordered:'yes');

Makelist.definekey('Make');

Makelist.definedone();

rc=Makelist.output(dataset: 'Manufacturer');

run;

Re: Does "definekey" enforce uniqueness in a hash object?

FreelanceReinh — Mon, 20 May 2019 13:32:44 GMT

Hi @Syntas_error,

@Syntas_error wrote:

Does this then mean that the definekey statement enforces uniqueness on those variable values, similarly to the PRIMARY KEY constraint?

No, it is the default multidata:'no' argument tag (in the parentheses after "Makelist" in the DECLARE statement) that makes Makelist a hash object with unique keys (which are then specified with the DEFINEKEY method).

(Also, note the typo in "dataest".)