Fill with @row_number based on group field in QGIS 3

Question

Is there a way to fill the table with a row number so that count will restart as new group (identified by a separate column) starts? As a result, "row number" column will restart the count when the new attribute value for a group is found, and will produce row number (1,2,3 etc.) record within this group.

This post on stack overflow seems to answer my question, but SQL in QGIS does not allow such statements (DENSE_RANK) (or it does but I don't know how?).

Answer 1

Do you need to use a field calculator? If not, the "Add autoincremental field" algorithm from the processing toolbox does what you want including the grouping.

---

References:

• How to Add Auto Increment Field in QGIS

Answer 2

Credits to @Gabriel De Luca from his answer https://gis.stackexchange.com/a/364575/99589

In the Field Calculator use the following expression:

array_find(array_agg($id, "group"), $id) + 1

where "group" is an attribute that will restart the count when a new group appears.

Answer 3

If you know the row number of the beginning of each group, you can create a group_row_number field in the Field Calculator with a simple case when statement.

In my example, I have groups A, B, C and D. Group A is rows 1-4. Group B is rows 5-8. Group C is rows 9-12. Group D is rows 13-17.

Field Calculator expression:

Result:

The generalized formula is:

case
  when "groupfieldname" = 'first group name' then @row_number
  when "groupfieldname" = 'second group name' then @row_number - [final_row_number_of_first_group]
  when "groupfieldname" = 'third group name' then @row_number - [final_row_number_of_second_group]
  ...
end

Answer 4

You can also do this with QGIS 3 field calculator more easy:

array_find(array_agg("GROUPFIELDNAME"||'_'||-$id,"GROUPFIELDNAME"),"GROUPFIELDNAME"||'_'||-$id)+1

Replace "GROUPFIELDNAME" with the name of your field you want to group by.