How to Remove Observation that do not look like an address

[SNG1]

Hi

I have a data set which is addresses due to the data being messy, I need to remove the address that look like a number;

For Example -

101Crawfordstreet

200miltonave

10-101adamsways

123

10

0

105R

I have already compressed the address to remove any blanks, I need to remove 0 10 123 102R etc. anything that does not make any sense...is there a way to do that.

[kiranv_]

I think you should put more logic for address but below code will work for you

 
data have;
input address $30.;
datalines;
101Crawfordstreet
200miltonave
10-101adamsways
123
10
0
105R
;

data want;
set have;
if anyalpha(address) = 0 then delete;
run;

[SuryaKiran · Accepted solution]

In addition to @kiranv_, I would add another condition.

data want;
set have;;
if anyalpha(address) = 0 or length(compress(address,,'ka'))<5   then delete;
run;

compress(address,,'ka') will keep only the alphabetic characters.

Likewise depending on your data you might add additional conditions that fits to your needs.

[SNG1]

Thanks Surya...your worked perfectly for my purpose much appreciated

[Reeza]

This article contains code that cleans up US addresses. I would suggest using it to parse the addresses. And any that don't parse as desired are deleted.

https://analytics.ncsu.edu/sesug/2008/CC-028.pdf

---

@SNG1 wrote:

Hi

I have a data set which is addresses due to the data being messy, I need to remove the address that look like a number;

For Example -

101Crawfordstreet

200miltonave

10-101adamsways

123

10

0

105R

I have already compressed the address to remove any blanks, I need to remove 0 10 123 102R etc. anything that does not make any sense...is there a way to do that.

---

[SNG1]

Thanks a Lot Everyone the solution provided worked from all by adding a few changes (i.e. filters) of my own...much appreciated.

Thanks!