topic Re: SPEDIS function in SAS Programming

SPEDIS function

sirivally — Mon, 01 Jul 2024 16:35:05 GMT

Hi,

I'm new to SAS. I'm trying to understand the behavior of SPEDIS function.

I ran below SAS code.

data new;
a = spedis('uzfiy', 'fuziy');
run;

According to my understanding from SAS documentation :

Here, three operations are done

Query | Keyword

uzfiy fuziy

Operation | Cost

u - f => firstrep 200

z - u => replace 100

f - z => replace 100

so, the result should be total cost / length of query => (200 + 100 + 100)/ 5 = 80.

But the actual result is 40.

can you please give me the actual operations done in this case ? And is there a way to get the source code of SPEDIS function?

Re: SPEDIS function

ChrisNZ — Tue, 02 Jul 2024 11:30:45 GMT

The function tries for operations with the lowest score.

Consider:

data new;
  a = spedis('ABC', 'XAX');  
run;

This has a cost of 300 (and a distance of 100):

Swap A to first position: cost=100

Replace the other 2 letters: cost =2x100

Replacing everything would cost 200+100+100=400.

Similarly in your case, there are swaps rather than replacements used in the process.

Letters u and z can be put in the correct position with a swap, and f has to be inserted. Cost = 2x50+100 = 200, for a distance of 200/5 = 40.

Re: SPEDIS function

rudfaden — Tue, 02 Jul 2024 12:27:17 GMT

if you swap f and u uzfiy you done and you get 200/5=40

Re: SPEDIS function

rudfaden — Tue, 02 Jul 2024 12:44:16 GMT

Sorry. All wrong.

Re: SPEDIS function

sirivally — Tue, 02 Jul 2024 14:51:27 GMT

Thank you for your response.

Just want to make sure if it is 4 swaps or 2 swaps + 1 insertion in my case ?

If 4 swaps:

uzfiy - zufiy - 50

zufiy - zfuiy - 50

zfuiy - fzuiy - 50

fzuiy - fuziy - 50

= 200

SPEDIS - 200/5 = 40

And is there a way to get the source code of SPEDIS function?

Re: SPEDIS function

SASKiwi — Tue, 02 Jul 2024 21:57:37 GMT

SAS does not usually provide source code for its software as it is proprietary.

Re: SPEDIS function

Kurt_Bremser — Wed, 03 Jul 2024 11:06:42 GMT

@sirivally wrote:

And is there a way to get the source code of SPEDIS function?

You have to ask SAS themselves for this, but I doubt they will provide it. SAS itself is not open source.

And it would only be helpful if you're good with C.

Re: SPEDIS function

ChrisNZ — Wed, 03 Jul 2024 10:56:21 GMT

Just want to make sure if it is 4 swaps or 2 swaps + 1 insertion in my case ?

You swap from the start string, not from another combination. Otherwise you could just move a letter along till the end.

Re: SPEDIS function

Kurt_Bremser — Wed, 03 Jul 2024 11:10:00 GMT

I still don't get how 2 swaps + 1 insert can work. After the swaps, you'd still have 5 characters, and the insert would result in 6 characters overall.

Could you please elaborate how the exact sequence 2 swaps + 1 insert works with the original text of the OP?

Re: SPEDIS function

ChrisNZ — Wed, 03 Jul 2024 12:26:49 GMT

I still don't get how 2 swaps + 1 insert can work

I meant 2 swaps + 1 replace, sorry about the confusion.