Conditional recode
Hi All,
Is there a way to recode specific string value based on other column values? for exmaple I have 1000 rows, I know that out of all those I want to recode only specfic products (in my case I have 2 products with the same name, and I want to make them distinct adding a group for them (but only for them). Something like on a screen below:
Thank you,
eM
Best Answers
-
emate Dataiku DSS Core Designer, Neuron 2020, Registered, Neuron 2021, Neuron 2022 Posts: 91 ✭✭✭✭✭✭Options
OK, nevermind....
if (Products=="A" && GROUP=="1", 'A', 'A1')
if (Products=="A" && GROUP=="2", 'A', 'A2')
*** OR maybe no.... when I used this formula in my 'real data' it recoded all values in Product Column to a A1.. hmm.
*** LAST UPDATE - sorry I already found the solution:
if(condition, then, else) so the syntax will be:
if(Group=="1", replace(Products, "A", "A1"),Products)
-
Hi,
To do this you will need to go through 2 visual recipes. The first step is to do a Window recipe to count the number of rows for a specific Product code.
Open a window recipe and set it up as below - Partitioning on your Product column and Count aggregation on your Product column (here it's Col1 in my example)
You will end up with a dataset that should look like this:
From there, open a Prepare recipe and use a Formula processor to create your Products after recode with a syntax like below:
Hope this helps
Answers
-
emate Dataiku DSS Core Designer, Neuron 2020, Registered, Neuron 2021, Neuron 2022 Posts: 91 ✭✭✭✭✭✭Options
Thank you for another way to approach this.
As for now, I think, since I know what specific product I am looking for, and that won't change I guess I can stick to my solution? Your way is much safer if I would not have idea how much products are like that, and when that can change over time.