# How to process multivariant questions in Orange

can you explain how to process multivariant questions in Orange? For example: I have sociological survey database. Among the others there is question:

How would you like to spend your weekend? (pick only 3):

1. At home
2. With friends
3. With family
4. At work

And here is database example:

Database logic: max three options to pick - 3 columns. Respondent answer's order doesnt matter. Is there a possibility to make frequency histogramm in Orange based on these three columns?

Since you say the order of picks doesn't matter, I think your data will be more easily manageable (in any tool, including Orange) if you first transform the database into a wide table of one-hot encoded features, with every option its own column:

id    weekend_home    weekend_friends    weekend_family    weekend_work
1     1               1                  1                 0
2     1               1                  0                 0
3     0               1                  0                 0
4     1               1                  0                 0
5     1               1                  1                 0

• Thx, and how can I process such kind of data in Orange? For example how can I create frequency diagram – Serg4356 Jul 2 '18 at 19:22
• I don't think you can do it easily in Orange. Perhaps with Python Script widget, but then it's all the same (but much easier) to use pandas. – K3---rnc Jul 2 '18 at 23:15

In orange there is a function "create class" (data module), there you can select the necessary answers and convert them into a dichotomous scale. You have a categorical scale, you need to create a more complex rule, perhaps with the calculation of a new variable.