[Solved] Converting Numerical Values to Binomial

neilduggan

Hi

Very new user here so apologies if this questions seems dumb!!

I have some survey data with answers to each question as follows:

Yes = 1
No = 2
Undecided = 3
Refused to answer = 4
etc

I want to remove all values not equal to 1 or 2 or set it to "N/A" and then set values = 1 to "True" and set values = 2 to "False"

At the moment, I'm using "Generate Attributes" to create new attributes called "answer_yes" (answer==1) and "answer_no" (answer==2)

I presume there is a much better way to do this?? Or is this way as simple as any??

Any help appreciated :-)

Thanks

Neil

neilduggan

One other thing - I can't remove the rows containing the values 3 or 4 as these occur on different lines each time.

Marco_Boeck

Hi,

does each row only contain the answer to a single question? Is the anwer column nominal? f so, you can use the Filter Examples operator to get ride of rows where the value is not 1 or 2. Then you can use Generate Attributes (on the very same answer column, the operator can overwrite existing values) with this expression:

if (Answer=="1", "true", "false")

otherwise you can do

if (Answer=="1", "true", if (Answer=="0", "false", NaN))

Regards,
Marco

neilduggan

Hi

Thanks for your reponse Marco.

Each row contains the answer the multiple questions (i.e. each column is a question and each row is a respondent)

Each answer is nominal (i.e. 1, 2, 3, 4, etc) although is viewed by RapidMiner as numerical.

I'll have a try when I get home and let you know how I got on

Thanks

neilduggan

Thanks Marco, that second piece of code did the trick