Sign up to take part
Registered users can ask their own questions, contribute to discussions, and be part of the Community!
Compare two datasets column by column without using code recipe
Added on December 27, 2023 10:20PM
Likes: 0
Replies: 1
yesitsmeoffical
Registered Posts: 15 ✭✭✭
How to create dataset 3 (key = 'ID', join method = 'left') to compare datasets 1 and 2 WITHOUT using code recipe, i.e low code no code only. Thanks.
Dataset 1:
| ID | Col A | Col B | Col C |
| 1 | AA | 11 | Z1 |
| 2 | BB | 100 | X2 |
| 3 | CC | 33 | Y5 |
Dataset 2:
| ID | Col A | Col B | Col C |
| 1 | AA | 11 | Z1 |
| 2 | BB | 98 | X2 |
| 3 | CC | 33 | YY555 |
| 4 | DD | 44 | T6 |
Dataset 3 (string columns return true/false, numeric columns return ratio):
| ID | Col A | Col B | Col C |
| 1 | TRUE | 100% | TRUE |
| 2 | TRUE | 98% | TRUE |
| 3 | TRUE | 100% | FALSE |
Operating system used: win 11
Best Answer
-
Sarina Dataiker, Dataiku DSS Core Designer, Dataiku DSS Adv Designer, Registered Posts: 319 DataikerHi @yesitsmeoffical
,
You can utilize the "pre-join computed columns" step within the join recipe to calculate your computed columns. From your description it sounds like you want to return TRUE if dataset 1 column A == dataset 2 column A and if dataset 1 column C == dataset 2 column C. And it looks like for column B you want to return dataset 2 col B / dataset 1 column B.
I created the following output dataset using your sample inputs and three computed columns:
The string computed columns have the formulas:
if(val("d1__Col A") == val("d2__Col A"), 'TRUE', 'FALSE') if(val('d1__Col C') == val('d2__Col C'), 'TRUE', 'FALSE')
The numeric computed column has the formula:100 * (val('d2__Col B') / val('d1__Col B'))
I'm attaching an export of the project so that you can test it out by importing into your DSS instance.
Thanks!
Sarina