How to calculate the attribute and replace the number and letter to the calculated value in rapidmi.

jabr

In the dataset, we have an attribute size that contains the size of the app like 12M, 45K, etc. I would like to calculate the 12M as 12*1000000 and replace the value as 12000000 and for 45K as 45*1000 and replace the value as 45000.

I have added a map operator and given the regular expression. But it not calculating in the Rapid miner. Please help  

Caperez

Hi @jabr, 

please find attached an example to convert it. 

Best

lionelderkrikor

Hi @jabr,

In attached file, you can find a working process which performs the requested transformation. (....inspired by @ceaperez solution ...

The trick here is to : 
 - replace "M" by "*1000000" (and "k" by "*1000")
 - split (using the Split operator) based on the caracter "*" pattern which produce 2 new attributes (size_1 and size_2)
- generate the requested attribute (New size) by multiplying "size_1 by "size_2"

You can remove the intermediate attributes  (Size_1 and Size_2) by adding and selecting these 2 attributes in the Select Attributes operator.

Hope this helps,

Regards,

Lionel 

Caperez

Hi @jabr, 
One way is to use Replace Operator to remove the k character using a regular expression [k]. Then you can use the Parse Numbers Operator to transform the polynominal attribute to numeric. Withe the numerical attribute you can create a new one with a simple calculation

Best

jabr

Thanks, @ceaperez for the solution. I can replace the K character. But if I remove the character then I won't be able to calculate it. All sizes 12M, 14K are given together in one column. If it is M then the number should be multiplied by 1000000 and if it is K then the number should be multiplied by 1000.  

Caperez

Hi @jabr, 

please find attached an example to convert it. 

Best

lionelderkrikor

Hi @jabr,

In attached file, you can find a working process which performs the requested transformation. (....inspired by @ceaperez solution ...

The trick here is to : 
 - replace "M" by "*1000000" (and "k" by "*1000")
 - split (using the Split operator) based on the caracter "*" pattern which produce 2 new attributes (size_1 and size_2)
- generate the requested attribute (New size) by multiplying "size_1 by "size_2"

You can remove the intermediate attributes  (Size_1 and Size_2) by adding and selecting these 2 attributes in the Select Attributes operator.

Hope this helps,

Regards,

Lionel 

jabr

Hi @ceaperez, 

Thanks a million. It is working perfectly

jabr

@lionelderkrikor Thanks a million