postgresql: create graph based on count of two different timestamp columns vs create_time - postgresql

Question

So I have a table with three columns:
create_time (date of table entry), process_time (date order was processed), report_time (date order was reported). Chronologically speaking, the order is always the following: process_time > report_time > create_time.
Both process_time and report_time can be different than create_time or themselves. But the main column I want to compare against is the create_time.
I would like to create a graph where the X column is the date of create_time and the Y column is a count of how many times that create_time date appears in the process_time or report_time columns. Not a count of process_time / report_time cells which have a value, but a count of the actual date.
Very simple example:
| create_time | process_time | report_time |
|-------------|--------------|-------------|
| 2019-02-01 | 2019-01-27 | 2019-01-28 |
| 2019-02-20 | 2019-02-20 | 2019-02-20 |
| 2019-02-26 | 2019-02-20 | 2019-02-25 |
In this example the graph would show a count of 0 for the first create_time date, since there are no process_time or report_time values that match that same date. For the second create_time it would show a count of 2 process_time and 1 report time and for the third one it would show a count of 0.
Hope this makes sense.

Answer 1

Creating the sample table:
CREATE TABLE example_table(create_time DATE, process_time DATE, report_time DATE);
INSERT INTO example_table(create_time, process_time, report_time)
VALUES ('2019-02-01', '2019-01-27', '2019-01-28'),
('2019-02-20', '2019-02-20', '2019-02-20'),
('2019-02-26', '2019-02-20', '2019-02-25');
The query that first selects all distinct create_time values and then calculates the number of appearances of that date in the process_time and report_time columns.
WITH create_dates AS (
SELECT DISTINCT create_time FROM example_table
)
SELECT * FROM create_dates cd
CROSS JOIN LATERAL (
SELECT
COUNT(*) FILTER (WHERE cd.create_time = et.process_time) as process_time_count,
COUNT(*) FILTER (WHERE cd.create_time = et.report_time) as report_time_count
FROM example_table et
) temp;
The result:
+------------+--------------------+-------------------+
| crete_time | process_time_count | report_time_count |
+------------+--------------------+-------------------+
| 2019-02-20 | 2 | 1 |
+------------+--------------------+-------------------+
| 2019-02-01 | 0 | 0 |
+------------+--------------------+-------------------+
| 2019-02-26 | 0 | 0 |
+------------+--------------------+-------------------+