Removing a variable conditionally

[Jackwangliang]

Dear all,

How can I remove observations that contain a specific character?

For example, remove observations contain E

ID    DX     Diag_Sequence

1      96         1

1      E96       1

1        87        2

1        E87      2

1       V87      3 

Best,

Liang  

[PaigeMiller · Accepted solution]

if find(dx,'E','i')>0 then delete;

[Jackwangliang]

Great! Thank you!

[Jackwangliang]

Thank you all for your great replies!



What if remove E that the Diag_Sequence is the same as previous observation, but do not remove E that has different Diag_Sequence number, for example,



ID DX Diag_Sequence

1 96 1

1 E96 1 (remove)

1 87 2

1 E87 2 (remove)

1 V87 3

1 E23 4 (do not remove)

1 12 5

1 15 6

1 E15 6 (remove)



I am a new SAS user, thank you very much for helping.



Liang

[Reeza]

It depends on how complicated it gets.

If its a single value you can chain logic together, such as 

if dx=:’E’ and dx ne ‘E23’ then delete;

But if the logic gets too complex you need a different approach. Do you have those lists of what to include/exclude in a data set or are they in your head? If you make them into a data set an entirely different approach is warranted, which may be easier overall.

[Jackwangliang]

Hi, Reeza. Thanks for your comment. 

Here is a screenshot of part of my dataset. Variable Diagnosis_Code may contain observations with E, for example, E9505.

For this E9505, if its Diagnosis_Sequence is the same as previous observations, then delete this E9505 observation. But if this Diagnosis_Sequence is not the same as the previous one, keep it. How can I code this? 

Best,

Liang

[Reeza]

@Jackwangliang thats not at all like your initial question. 

Please post data as text, I won't type out your data, and what you would expect as output. Since this is so different from your original question, I would suggest you start a new thread. 

---

@Jackwangliang wrote:

Hi, Reeza. Thanks for your comment. 

Here is a screenshot of part of my dataset. Variable Diagnosis_Code may contain observations with E, for example, E9505.

For this E9505, if its Diagnosis_Sequence is the same as previous observations, then delete this E9505 observation. But if this Diagnosis_Sequence is not the same as the previous one, keep it. How can I code this? 

Best,

Liang

---

[Jackwangliang]

How can I post my data as text here? Thanks.

[Reeza]

https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-dat...

[mbuchecker]

proc sql;

delete from table_name

where DX contains 'E';

[Reeza]

If dx =: ‘E’ then delete; *removes all dx that start with E;
If find(dx, ‘E’, ‘i’)>0 then delete;*removes all dx that contain an E;

Use find to search for an E, there’s a possibility I have the parameters backwards in the FIND() function so you should verify that. 

Or use =: which checks if the first characters are the same. 

---

@Jackwangliang wrote:

Dear all,

How can I remove observations that contain a specific character?

For example, remove observations contain E

ID    DX     Diag_Sequence

1      96         1

1      E96       1

1        87        2

1        E87      2

1       V87      3 

Best,

Liang  

---

[Jackwangliang]

Thank you all for your great replies! 

What if remove E that the Diag_Sequence is the same as previous observation, but do not remove E that has different Diag_Sequence number, for example,

ID    DX     Diag_Sequence

1      96         1

1      E96       1     (remove)

1        87        2

1        E87      2    (remove)

1       V87      3 

1      E23       4     (do not remove)

1      12          5

1      15        6

1      E15     6      (remove)

I am a new SAS user, thank you very much for helping.

Liang