Postgres: Unable to determine percent of successful events ending in a completed trip - sql

Question

SQL Gurus,
I'm trying to solve this challenging problem as I'm practicing my SQL skills, however I'm stuck and would appreciate if someone could help.
A signup is defined as an event labelled ‘sign_up_success’ within the events table. For each city (‘A’ and ‘B’) and each day of the week, determine the percentage of signups in the first week of 2016 that resulted in completed a trip within 10 hours of the sign up date.
Table Name: trips
Column Name Datatype
id integer
client_id integer (Foreign keyed to
events.rider_id)
driver_id integer
city_id Integer (Foreign keyed to
cities.city_id)
client_rating integer
driver_rating integer
request_at Timestamp with timezone
predicted_eta Integer
actual_eta Integer
status Enum(‘completed’,
‘cancelled_by_driver’, ‘cancelled_by_client’)
Table Name: cities
Column Name Datatype
city_id integer
city_name string
Table Name: events
Column Name Datatype
device_id integer
rider_id integer
city_id integer
event_name Enum(‘sign_up_success’, ‘attempted_sign_up’,
‘sign_up_failure’)
_ts Timestamp with timezone
Tried something on this lines, however its no where near the expected answer:
SELECT *
FROM trips AS trips
LEFT JOIN cities AS cities ON trips.city_id = cities.city_id
LEFT JOIN events AS events ON events.client_id = events.rider_id
WHERE events.event_name = "sign_up_success"
AND Convert(datetime, trips.request_at') <= Convert(datetime, '2016-01-
07' )
AND DATEDIFF(d, Convert(datetime, events._ts), Convert(datetime,
trips.request_at)) < 7 days
AND events.status = "completed
Desired Results look like below:
Monday A x%
Monday B y%
Tuesday A z%
Tuesday A p%
Can someone please help.

Answer 1

First of all, I assume that "trips"."city_id" is mandatory, so I use INNER JOIN instead of LEFT JOIN when joining with cities.
Then, to specify string constants, you need to use single quotes.
There are some other changes in the query -- hope you'll notice them yourself.
Also, the query might fail, since I didn't run it actually (you didn't provide boilerplate SQL unfortunately).
date_trunc() function with 'week' first parameter converts your timestamp to "first day of the corresponding week, time 00:00:00", based on your current timezone settings (see https://www.postgresql.org/docs/current/static/functions-datetime.html).
I used GROUP BY on that value and second "layer" of grouping was city ID.
Also, I used "filter (where ...)" next to count() -- it allows to count only desired rows.
Finally, I used CTE to improve the query's structure and readability.
Let me know if it fails, I'll fix it. In general, this approach must work.
with data as (
select
left(date_trunc('week', t.request_at)::text, 10) as period,
c.city_id,
count(distinct t.id) as trips_count,
count(*) filter (
where
e.event_name = 'sign_up_success'
and e._ts < t.request_at + interval '10 hour'
) as successes_count
from trips as t
join cities as c on t.city_id = c.city_id
left join events as e on t.client_id = e.rider_id and e._ts
where
t.request_at between '2016-01-01' and '2016-01-08'
group by 1, 2
)
select
*,
round(100 * success_count::numeric / trips_count, 2)::text || '%' as ratio_percent
from data
order by period, city_id
;