Need help converting multiple variables from Character to numeric

[RaginiS]

Hi, I am super new to SAS programming and need help,

I have a large SAS dataset of property prices information and I need to convert about 3 columns from character to numeric by assigning values 1,2.3….

1. for one column for HeatingAC – I need to put Yes-1 and No as 0
2. For ExternalQuality and KitchenQuality- I need to use 5- for Excellent, 4-Good, 3-Average, 2-Fair, 1-Poor.
3. Dwellingtype- 1-Singlefamily, 2-Twofamily, 3-Duplex and so on;

I tried doing all of this is one temporary dataset but it shows errors, please help. Thanks!

[Reeza]

---

@RaginiS wrote:

Hi, I am super new to SAS programming and need help,

I have a large SAS dataset of property prices information and I need to convert about 3 columns from character to numeric by assigning values 1,2.3….

1. for one column for HeatingAC – I need to put Yes-1 and No as 0
2. For ExternalQuality and KitchenQuality- I need to use 5- for Excellent, 4-Good, 3-Average, 2-Fair, 1-Poor.
3. Dwellingtype- 1-Singlefamily, 2-Twofamily, 3-Duplex and so on;

I tried doing all of this is one temporary dataset but it shows errors, please help. Thanks!

---

Please show what you tried, your code and the log. You likely have a small issue that's easy to correct. 

Video tutorial is here. Ensure you create NEW variables to hold the new values as well, you cannot change the type of a variable on in the same data step. 

https://video.sas.com/detail/videos/sas-analytics-u/video/4573023399001/creating-a-new-column-in-sas...

[RaginiS]

Hi, This is what I have so far and it does not work-

libname Home '/folders/myfolders/home';

data test1;
set home.realestate (rename=(HeatingAC=oldvar));
if oldvar = 'N' then HeatingAC = 0;
else HeatingAC= 1;
drop oldvar;

data test1;
set home.realestate ;
array clean{*} $ _character_;

do i=1 to dim(clean);

if clean{i}='SingleFamily' then
clean{i}= 1;

if clean{i}='TwoFamily' then
clean{i}=2;

if clean{i}='Duplex' then
clean{i}=3;

if clean{i}='Townhome' then
clean{i}=4;

if clean{i}='Condo' then
clean{i}=5;
end;

drop i;
RUN;

proc print data=test1;
run;

[Reeza · Accepted solution]

Either combine your steps or make sure they don't all start from the original file (realestate). If they do, they're unique files and the changes made don't carry over. I think this is what you were trying to  do, changes are in red.

data test1;
set home.realestate (rename=(HeatingAC=oldvar));
if oldvar = 'N' then HeatingAC = 0;
else HeatingAC= 1;
drop oldvar;
run;


data test2;
set test1 ;
array clean{*} $ _character_;  *are you sure you want all character variables here;

do i=1 to dim(clean);

if clean{i}='SingleFamily' then clean{i}= 1;
else if clean{i}='TwoFamily' then clean{i}=2;
else if clean{i}='Duplex' then clean{i}=3;
else if clean{i}='Townhome' then clean{i}=4;
else if clean{i}='Condo' then clean{i}=5;

end;

drop i;
RUN;

proc print data=test2;
run;

Let me know if that gets you closer. I'm not sure that using _character_ here makes sense, it will use all character variables including an ID or anything else. 

---

@RaginiS wrote:

Hi, This is what I have so far and it does not work-

libname Home '/folders/myfolders/home';

data test1;
set home.realestate (rename=(HeatingAC=oldvar));
if oldvar = 'N' then HeatingAC = 0;
else HeatingAC= 1;
drop oldvar;

data test1;
set home.realestate ;
array clean{*} $ _character_;

do i=1 to dim(clean);

if clean{i}='SingleFamily' then
clean{i}= 1;

if clean{i}='TwoFamily' then
clean{i}=2;

if clean{i}='Duplex' then
clean{i}=3;

if clean{i}='Townhome' then
clean{i}=4;

if clean{i}='Condo' then
clean{i}=5;
end;

drop i;
RUN;

proc print data=test1;
run;

---

[RaginiS]

Thank you, that worked 

[RaginiS]

Quick question, I was able to covert the character variables to numeric, I can see it when I run proc print but when I run proc contents it still shows as Char, what did I not do?

[Reeza]

You cannot change variable types on the fly. You need to assign it to new variables, so just duplicate your array with a set of new names and those can be numeric.

[RaginiS]

Thanks!