Calculate Status Chages and time in between status - sql

Question

I have a table as shown below. I have column that tracks an application and application status changes over time. I track the time it changes the status in the date column. The table is sorted by application and date of status change from oldest to newest.
+--------+-----------+--------+------------------+
| app_id | status_id | row_no | date |
+--------+-----------+--------+------------------+
| 1 | a | 10 | 2016-10-04 21:35 |
| 1 | b | 11 | 2016-10-12 21:50 |
| 1 | c | 12 | 2016-10-25 20:40 |
| 1 | d | 13 | 2016-10-26 16:10 |
| 1 | e | 14 | 2016-10-26 16:10 |
| 2 | a | 20 | 2016-09-15 1:26 |
| 2 | c | 21 | 2016-09-15 21:32 |
| 2 | d | 22 | 2016-09-16 21:51 |
| 2 | e | 23 | 2016-09-16 21:51 |
| 2 | f | 24 | 2016-09-20 22:55 |
| 2 | g | 25 | 2016-10-20 22:46 |
| 2 | g | 26 | 2016-10-20 22:46 |
+--------+-----------+--------+------------------+
I am trying to achieve how much time an application is spending before it reached the final state. Below is a sample of the table That i am trying to build in Sql. For every status i am trying to capture the next status. The previous status column shows the status in that row, while next status shows the next status in the next row for that application. If the application is at its last status then the next status is marked as Last. Next I calculate the time between status by calculating the hours differences between two dates. I would really appreciate if you can tell me how to achieve this functionality in Sql. Thank you in advance.
+--------+-----------+--------+------------------+-----------------+-------------+--------------+
| app_id | status_id | row_no | date | previous status | next status | time between |
+--------+-----------+--------+------------------+-----------------+-------------+--------------+
| 1 | a | 10 | 2016-10-04 21:35 | a | b | 192.2333333 |
| 1 | b | 11 | 2016-10-12 21:50 | b | c | 334.8333333 |
| 1 | c | 12 | 2016-10-25 20:40 | c | d | 43.48333333 |
| 1 | d | 13 | 2016-10-26 16:10 | d | e | 0 |
| 1 | e | 14 | 2016-10-26 16:10 | e | Last | Last |
| 2 | a | 20 | 2016-09-15 1:26 | a | c | 20.08333333 |
| 2 | c | 21 | 2016-09-15 21:32 | c | d | 24.31666667 |
| 2 | d | 22 | 2016-09-16 21:51 | d | e | 0 |
| 2 | e | 23 | 2016-09-16 21:51 | e | f | 97.06666667 |
| 2 | f | 24 | 2016-09-20 22:55 | f | g | 743.8333333 |
| 2 | g | 25 | 2016-10-20 22:46 | g | g | 0 |
| 2 | g | 26 | 2016-10-20 22:46 | g | Last | Last |
+--------+-----------+--------+------------------+-----------------+-------------+--------------+

Answer 1

It's a bit messy, but if you have a known amount of status_id's, you can try making each one in a "with" clause and joining them all together at the end on app_id. And then making a final table which calculates the steps between A and B, between B and C, etc. This wouldn't produce a table like the one you made, however. But it should get all the time differences.
with A_table as (
select
app_id,
date A_status
where status_id = 'a'
)
, B_table (
select
app_id,
date B_status
where status_id = 'b'
)
--MORE STATUS TABLE HERE
, combined_table (
select
a.app_id,
a.A_status,
b.B_status,
--MORE STATUS DATES HERE
from A_table a
left outer join B_table b on a.app_id = b.app_id
--LEFT OUTER JOIN MORE STATUS TABLES ON A_TABLE HERE
--YOU'RE MAKING ONE TABLE WITH EACH APP_ID ON ONE ROW WITH ALL TIME STAMPS
)
select
*,
B_status - A_status A_B
--MORE TIME SUBTRACTIONS HERE
--SINCE YOU'VE OUTER JOINED ABOVE, YOU'LL HAVE COLUMNS FOR ALL POSSIBLE
--STATUS STEPS AND THOSE WHICH DIDN'T HAVE THAT STEP WILL BE NULL
from combined
It's kind of clunky but with a fixed amount of status steps, should get the job done. It doesn't account for which step is the "last step" though. I don't know how important that is. You can always write a case statement that looks at the next step to see if it was null. What you want might be able to be achieved with loops but I've never used those.
Also note that if you have duplicate rows of app_id, status_id and date, like the last two rows in your sample table, you'd need to sort that out in the with tables some how, like only taking the first row, or ranking them.

Answer 2

Using Sql Lead and Lag function we can achive this. Here is the query:
select app_id, status as prev_status, date as prev_date, lead(status) over (partition by app_id order by date) as next_status, lead(date) over (partition by app_id order by date) as next_date from table