Summary of Workflow History over Time - sql-server

Question

I am collecting data using a web formular and a workflow with four different status.
Started
Review by Admin
Back to User
Finished
Status 2 and 3 are repeated as long as it is needed. The application logs the history of this process and I would like to be able so see the development of the workflow over time.
The history is stored in an MS SQL Table and I can rearange it to look like this.
In the example there are two workflows (id 1 and 2) and over time the formular goes back and forth untill it is finished. I can order this by time
to derive the result I would like to have.
I dont need to have this result in wide format as long as the counting in long is correct.
I am struggling to find a way to avoid double counting status 2 and 3 and also I dont want to count status 1 later on.
I am rather looking for a strategy to obtain my desired results either in SQL or Power Query.
I tried different grouping, dense_rank and other transformations in Power Query.

Answer 1

Try this in PowerQuery, assumes already date sorted
let Source = Excel.CurrentWorkbook(){[Name="Table1"]}[Content],
#"Changed Type" = Table.TransformColumnTypes(Source,{{"Datum", type datetime}}),
#"Added Custom" = Table.AddColumn(#"Changed Type", "zStarted", each if [Status]="Started" then 1 else if [Status]="Review by Admin" then -1 else 0),
#"Added Custom1" = Table.AddColumn(#"Added Custom", "zReview", each if [Status]="Review by Admin" then 1 else if [Status]="Back to User" then -1 else if [Status]="Finished" then -1 else 0),
#"Added Custom2" = Table.AddColumn(#"Added Custom1", "zBackA", each if [Status]="Back to User" then 1 else null),
#"Filled Down" = Table.FillDown(#"Added Custom2",{"zBackA"}),
#"Added Custom3" = Table.AddColumn(#"Filled Down", "zBackB", each if [zBackA]=null then 0 else if [Status]="Back to User" then 1 else if [Status]="Review by Admin" then -1 else 0),
#"Added Custom4" = Table.AddColumn(#"Added Custom3", "zFinished", each if [Status]="Finished" then 1 else 0),
#"Added Index" = Table.AddIndexColumn(#"Added Custom4", "Index", 0, 1, Int64.Type),
// cumulative sum, converting to zero if <0
#"Cum1" = Table.AddColumn(#"Added Index", "Started", each List.Max(List.Combine({{0}, {List.Sum(List.FirstN(#"Added Index"[zStarted],[Index]+1))}}))),
#"Cum2" = Table.AddColumn(#"Cum1", "Review By Admin", each List.Max(List.Combine({{0}, {List.Sum(List.FirstN(#"Added Index"[zReview],[Index]+1))}}))),
#"Cum3" = Table.AddColumn(#"Cum2", "Back to User", each List.Max(List.Combine({{0}, {List.Sum(List.FirstN(#"Added Index"[zBackB],[Index]+1))}}))),
#"Cum4" = Table.AddColumn(#"Cum3", "Finished", each List.Max(List.Combine({{0}, {List.Sum(List.FirstN(#"Added Index"[zFinished],[Index]+1))}}))),
#"Removed Columns" = Table.RemoveColumns(Cum4,{"ID","zStarted", "zReview", "zBackA", "zBackB", "zFinished", "Index","Status"})
in #"Removed Columns"
ID
Datum
Status
1
06/24/22
Started
2
06/25/22
Started
1
06/26/22
Review by Admin
2
06/27/22
Review by Admin
1
06/28/22
Back to User
1
06/29/22
Review by Admin
2
06/30/22
Back to User
1
07/01/22
Back to User
1
07/02/22
Review by Admin
2
07/03/22
Review by Admin
1
07/04/22
Finished
2
07/05/22
Back to User
2
07/06/22
Review by Admin
2
07/07/22
Finished

Answer 2

Thanks to #horseyride i came up with my own solution using case and a similar approach within SQL
I follow a three step appraoch to avoid the double counting of changes in status by using the lagged status (sql columne action_nam)
With the lagged status I can then unambigously define the changes of the status and take the cumulutive sum.
I mark the first answer as accepted as this one helped me to find my own solution. I leave my solution with the Germany Status names just for completeness
Select top 600 created, form_uuid, Cum_Erstellt, Cum_Review, Cum_Korrektur, Cum_Fertig from (
Select *,
sum(Erstellt) over(Partition by form_uuid order by form_uuid, created) as Cum_Erstellt,
sum(Review) over(Partition by form_uuid order by form_uuid, created) as Cum_Review,
sum(Korrektur) over(Partition by form_uuid order by form_uuid, created) as Cum_Korrektur,
sum(Fertig) over(Partition by form_uuid order by form_uuid, created) as Cum_Fertig,
Erstellt+Review+Korrektur+Fertig as Vergleich
from (
Select created, form_uuid, encoded_key, action_name, lag_action,
CASE WHEN action_name = 'Datensatz erstellt und zur Bearbeitung an Datenlieferanten übergeben' THEN 1
When action_name = 'Datensatz zur Prüfung an Datensammler' and
lag_action='Datensatz erstellt und zur Bearbeitung an Datenlieferanten übergeben' THEN -1
Else 0 END AS Erstellt,
CASE WHEN action_name = 'Datensatz zur Prüfung an Datensammler' THEN 1
When action_name= 'Datensatz zur Korrektur an Datenlieferanten' Then -1
when Action_name= 'Datensatz erneut Öffnen' then 1
When action_name='Datensatz durch Datensammler bestätigt' Then -1
Else 0 END AS Review,
CASE
WHEN action_name = 'Datensatz zur Prüfung an Datensammler' and
lag_action!='Datensatz erstellt und zur Bearbeitung an Datenlieferanten übergeben' THEN -1
When action_name= 'Datensatz zur Korrektur an Datenlieferanten' Then 1
Else 0 END AS Korrektur,
CASE WHEN action_name = 'Datensatz durch Datensammler bestätigt' THEN 1
When action_name= 'Datensatz erneut Öffnen' Then -1
Else 0 END AS Fertig
from (
Select created, form_uuid, encoded_key, action_name, lag(action_name,1,NULL)
over( order by encoded_key, form_uuid, created) as lag_action
from process_instance_hist ) lag_table
) change_table
) cum_table
order by created, encoded_key
--group by created, form_uuid