SQL Calculate daily increment with Date and ID on multiple variables - sql

Question

I have a table with the following structure:
+----------+-----+-------+----------+
| Date | ID | Likes | Comments |
+----------+-----+-------+----------+
| 1/1/2018 | AAA | 70 | 90 |
| 1/2/2018 | AAA | 80 | 110 |
| 1/1/2018 | BBB | 60 | 5 |
| 1/2/2018 | BBB | 90 | 6 |
+----------+-----+-------+----------+
For each day and ID, I need to calculate the incremental value. The desired output should something like this: desired output
+----------+-----+-------+----------+-------------+----------------+
| Date | ID | Likes | Comments | daily_likes | daily_comments |
+----------+-----+-------+----------+-------------+----------------+
| 1/1/2018 | AAA | 70 | 90 | 70 | 90 |
| 1/2/2018 | AAA | 80 | 110 | 10 | 20 |
| 1/1/2018 | BBB | 60 | 5 | 60 | 5 |
| 1/2/2018 | BBB | 90 | 6 | 30 | 1 |
+----------+-----+-------+----------+-------------+----------------+
I have tried this code but it keeps aggregating the daily_likes or daily_comments together.
SELECT
"date",
"created_time",
("likes_count" - LAG("likes_count", 1) OVER (ORDER BY "date")) AS "daily_likes",
("comments_count" - LAG("comments_count", 1) OVER (ORDER BY "date")) AS "daily_comments",
"id",
"likes_count",
"comments_count",
"user_username"
FROM
"blablabla"
GROUP BY
1,
"id",
"likes_count",
"comments_count",
"user_username",
"created_time"
ORDER BY
1 DESC;

Answer 1

The table structure you postet doesn't match the query. And I'm wondering why you group at all, let alone on nearly every column and a literal...
A query, that would produce the desired output you posted from a table like you posted would be:
SELECT "date",
"id",
"likes",
"comments",
"likes"
- lag("likes",
1,
0) OVER (PARTITION BY "id"
ORDER BY "date") "daily_likes",
"comments"
- lag("comments",
1,
0) OVER (PARTITION BY "id"
ORDER BY "date") "daily_comments"
FROM "elbat"
ORDER BY "date",
"id";
SQL Fiddle
There is no need to group at all.
(However note, to show the daily increase, this requires the source table to have data for every day for every user. If you got gaps, you'd need to join all days (in the respective range) for all users to fill them.