Split the table in DSS into separate test and train set table

Question

Hi,

I want to split the data table in the workflow into train and test set with split done randomly. How can we do that? I tried using "split" recipe but i didnt get the desired output.. Can any one help me with this ?

 

Sam

Comments

Hi Sam, can you provide more information on what you tried, what was the actual output, and what was the desired output?

---

I have uploaded a CSV table in DSS and now i want to split the table into 2 parts (Train test with random seeds) in the flow.  I thought that "split" recipe would do that but it did not have the option of splitting it randomly. Can you please let me know how we can split the table into 2 parts

I am also having problem with this link
http://answers.dataiku.com/1367/how-can-we-export-the-predicted-data-table-of-a-model-to-csv

Thank you

Answer 1

These settings will let you achieve what you desire:

- define a custom variable "rand()"

- add two filters on that variable, one <0.8, the other >=0.8