Hi all, I'm evaluating to use RM in my project also to perfom typical ETL operations.
I've already found in this forum that Rapid Miner main purpose is not ETL (e.g.
http://rapid-i.com/rapidforum/index.php/topic,986.0.html) , but I’d like to manage to fulfill my ETL requirements to use after all the power of Rapid Miner in Data Mining.
So in below I express the solution I found for some ETL operation … thanks in advance if someone has a more efficent suggestion to make the same thing.
My example has the Iris Data as input and a MySQL table as output.
1. APPEND ⇒
The aim is to append new record in the table (with a primary key): but the input set has also record the can violate the referential integrity
Operator WriteDB:
overwrite and overwrite first, append then (by the way, what is the difference??) ⇒the result is that the table is deleted and filled with new data ⇒ no good because I lose data initially stored in the table
append ⇒ error, due to refrerential integrity violation
The solution found is to filter input data and to append only the non-violating data by using directly SQL (by Execute SQL operator).
Is there a better solution for this problem?
2. MAPPING
The aim is to trasfer (append) records from a source table and target table with different fileld names.
How to bulid a mapping to append correctly the data, using in this case Operator WriteDB, method Append (the hypotesisi in no referential integrity problem in this case)?
Thanks in advance.
