Special Character - Example Value

Riitu

When special characters are on rows (example value) are not allowed in RapidMiner attribute. I spent hours to debug this but no luck! Rapid Miner doesn't process any attribute names with  'double quotes' (").  I found that "#", "-", "/", "(", ")" work OK.

 Anyways I can solve this issue?

Thanks...

[Deleted User]

@Riitu

I had the same problem. If you change those characters with alphabet you can solve it

Regards
mbs

Riitu

Thanks Mbs! Yes,  I can change in the dataset but I was looking for an operator which can handle such special characters.

Regards, 
Riitu 

[Deleted User]

I am not sure about operator may be others can help you more.

But there is a very useful document for operators and here is the link of that: 

https://docs.rapidminer.com/latest/studio/operators/rapidminer-studio-operator-reference.pdf

you can take a look on that may be you can find an operator for that.

All the best
mbs

hbajpai

@Riitu
You can use Replace operator to handle any special characters. Let's say you want to replace all special characters with a blank space you can configure the replace what with regex of all the special characters [-!"#$%&'()*+,./:;<=>?@\[\\\]_`{|}~] and leave replace by as blank. In the operator, you can filter attributes on the basis of your requirement. There are other ways to handle any exceptions, can you elaborate more on the problem.

Riitu

Am sharing the table ..named as Track. When you open this track.csv it shows total records as 3503 but on Rapidminer as 3498 (so 3503-3497=5). 5 rows are having special characters :

109

2918

3027

3254

3412
Bcos of above 5 rows are not imported on Rapidminer. Hence can't use Replace operator. I had to replace on Excel sheet

kayman

One (but clumsy) way to deal with this is to import your data without using row names. 
So your recordset will be like att_1 to att_n on import, which is the first step.

Next you use find and replace on row 1 (having your final row names) and replace \W with nothing. So any non word character (\W) will be removed in essence. This will leave you with all your 'to be titles' but without weird characters. You could also replace bu a space or underscore, up to you in the end.

As a final step you use the 'Rename by Example Values' operator and set record 1, this will rename all of the original attributes with your cleaned data on row 1.

[Deleted User]

@Riitu
Also there is an answer here in this link

https://community.rapidminer.com/discussion/56398/replace-hyphen#latest

All the best
mbs