counterweight/data-dwh-dbt-project
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merged PR 3187: Move deal lifecycle related models to cross

Browse source 
# Description

Moves from intermediate/core to intermediate/cross the following models:
- `int_core__mtd_deal_lifecycle`
- `int_core__mtd_deal_metrics`

to their equivalents:
- `int_mtd_deal_lifecycle`
- `int_mtd_deal_metrics`

This also changes the schema entries, from core to cross, including changing the name of the model in the entry.
This also changes the dependencies, namely in `int_mtd_deal_metrics`, `int_mtd_vs_previous_year_metrics` and `int_monthly_aggregated_metrics_history_by_deal`.
This does NOT aim to alter the logic of the lifecycle in any case; it will be done in a separated PR.

Runs correctly end-to-end. We might need to drop the old models from production manually.

# Checklist

- [X] The edited models and dependants run properly with production data.
- [X] The edited models are sufficiently documented.
- [X] The edited models contain PK tests, and I've ran and passed them.
- [X] I have checked for DRY opportunities with other models and docs.
- [X] I've picked the right materialization for the affected models.

# Other

- [ ] Check if a full-refresh is required after this PR is merged.

Related work items: #22689
This commit is contained in:
Oriol Roqué Paniagua 2024-10-16 11:56:49 +00:00
parent c0e411ee12
commit 004616bb79
6 changed files with 136 additions and 143 deletions
Download patch file Download diff file Expand all files Collapse all files

160
models/intermediate/core/int_core__mtd_deal_lifecycle.sql
View file

@ -1,160 +0,0 @@
/*
This model provides Month-To-Date (MTD) deal lifecycle based on booking metrics.
Assumes a host will have a deal being filled.
*/
{{ config(materialized="table", unique_key=["date", "id_deal"]) }}
with
int_core__bookings as (select * from {{ ref("int_core__bookings") }}),
int_core__unified_user as (select * from {{ ref("int_core__unified_user") }}),
int_dates_mtd as (select * from {{ ref("int_dates_mtd") }}),
booked_days_per_host as (
select
id_user_host,
created_date_utc,
lag(created_date_utc, 1) over (
partition by id_user_host order by created_date_utc asc
) as previous_booked_date
from int_core__bookings
group by id_user_host, created_date_utc
),
deal_historic_booking_dates as (
select
d.date,
h.id_deal,
min(h.created_date_utc) as creation_date_utc,
min(b.created_date_utc) as first_time_booked_date_utc,
max(b.created_date_utc) as last_time_booked_date_utc,
max(b.previous_booked_date) as second_to_last_time_booked_date_utc
from int_dates_mtd d
inner join int_core__unified_user h on d.date >= h.created_date_utc
left join
booked_days_per_host b
on h.id_user = b.id_user_host
and d.date >= b.created_date_utc
where h.id_deal is not null
group by d.date, h.id_deal
),
deal_historic_booking_features as (
select
hhbf.date,
hhbf.id_deal,
hhbf.creation_date_utc,
hhbf.first_time_booked_date_utc,
hhbf.last_time_booked_date_utc,
hhbf.second_to_last_time_booked_date_utc,
case
when hhbf.first_time_booked_date_utc is not null then true else false
end as deal_has_at_least_one_booking,
case
when
date_trunc('month', hhbf.creation_date_utc)
= date_trunc('month', hhbf.date)
then true
else false
end as deal_was_created_this_month,
case
when
date_trunc('month', hhbf.first_time_booked_date_utc)
= date_trunc('month', hhbf.last_time_booked_date_utc)
then true
else false
end as deal_has_first_booking,
case
when
date_trunc('month', hhbf.last_time_booked_date_utc)
= date_trunc('month', hhbf.date)
then true
else false
end as has_been_booked_within_current_month,
case
when
date_trunc('month', hhbf.last_time_booked_date_utc)
+ interval '6 months'
> date_trunc('month', hhbf.date)
then true
else false
end as has_been_booked_within_last_6_months,
case
when
date_trunc('month', hhbf.last_time_booked_date_utc)
+ interval '12 months'
> date_trunc('month', hhbf.date)
then true
else false
end as has_been_booked_within_last_12_months,
case
when
date_trunc('month', hhbf.last_time_booked_date_utc)
+ interval '12 months'
= date_trunc('month', hhbf.date)
then true
else false
end as last_booking_was_12_months_ago,
case
when
date_trunc('month', hhbf.second_to_last_time_booked_date_utc)
+ interval '12 months'
< date_trunc('month', hhbf.last_time_booked_date_utc)
then true
else false
end as had_previous_booking_more_than_12_months_before_the_last
from deal_historic_booking_dates hhbf
)
select
date,
id_deal,
creation_date_utc,
first_time_booked_date_utc,
last_time_booked_date_utc,
second_to_last_time_booked_date_utc,
case
-- 01-New: The deal has been created this month and has not had any booking
when not deal_has_at_least_one_booking and deal_was_created_this_month
then '01-New'
-- 02-Never Booked: The deal has been created before this month and has not
-- had any booking
when not deal_has_at_least_one_booking and not deal_was_created_this_month
then '02-Never Booked'
-- 03-First Time Booked: The deal has been booked for the first time and it
-- has been created this month
when deal_has_first_booking and has_been_booked_within_current_month
then '03-First Time Booked'
-- 04-Active: The deal has had at least 1 booking in its history and it's
-- been less than 12 months since the last booking
-- and is not reactivated and is not FTB
when
deal_has_at_least_one_booking
and has_been_booked_within_last_12_months
-- not reactivated
and not (
had_previous_booking_more_than_12_months_before_the_last
and has_been_booked_within_current_month
)
-- not FTB
and not (deal_has_first_booking and has_been_booked_within_current_month)
then '04-Active'
-- 05-Churning: The deal has been booked at least once and it's been 12
-- months since the last booking
when deal_has_at_least_one_booking and last_booking_was_12_months_ago
then '05-Churning'
-- 06-Inactive: The deal has been booked at least once and it's been more
-- than 12 months since the last booking
when
deal_has_at_least_one_booking
and not has_been_booked_within_last_12_months
and not last_booking_was_12_months_ago
then '06-Inactive'
-- 07-Reactivated: The deal was churned/inactive but has had a new booking
-- this month
when
had_previous_booking_more_than_12_months_before_the_last
and has_been_booked_within_current_month
then '07-Reactivated'
else null
end as deal_lifecycle_state,
has_been_booked_within_current_month,
has_been_booked_within_last_6_months,
has_been_booked_within_last_12_months
from deal_historic_booking_features

128
models/intermediate/core/int_core__mtd_deal_metrics.sql
View file

@ -1,128 +0,0 @@
/*
This model provides Month-To-Date (MTD) based on Deal metrics.
*/
{% set dimensions = get_kpi_dimensions() %}
{{ config(materialized="table", unique_key=["date", "dimension", "dimension_value"]) }}
with
int_core__mtd_deal_lifecycle as (
select * from {{ ref("int_core__mtd_deal_lifecycle") }}
),
int_dates_mtd_by_dimension as (
select * from {{ ref("int_dates_mtd_by_dimension") }}
),
int_core__mtd_accommodation_segmentation as (
select * from {{ ref("int_core__mtd_accommodation_segmentation") }}
),
int_core__deal as (
select * from {{ ref("int_core__deal") }}
),
deals_metric_aggregation_per_date as (
{% for dimension in dimensions %}
select
al.date,
{{ dimension.dimension }} as dimension,
{{ dimension.dimension_value }} as dimension_value,
count(
distinct case
when al.deal_lifecycle_state = '01-New'
then al.id_deal
else null
end
) as new_deals,
count(
distinct case
when al.deal_lifecycle_state = '02-Never Booked'
then al.id_deal
else null
end
) as never_booked_deals,
count(
distinct case
when al.deal_lifecycle_state = '03-First Time Booked'
then al.id_deal
else null
end
) as first_time_booked_deals,
count(
distinct case
when al.deal_lifecycle_state = '04-Active'
then al.id_deal
else null
end
) as active_deals,
count(
distinct case
when al.deal_lifecycle_state = '05-Churning'
then al.id_deal
else null
end
) as churning_deals,
count(
distinct case
when al.deal_lifecycle_state = '06-Inactive'
then al.id_deal
else null
end
) as inactive_deals,
count(
distinct case
when al.deal_lifecycle_state = '07-Reactivated'
then al.id_deal
else null
end
) as reactivated_deals,
sum(
case when has_been_booked_within_current_month then 1 else 0 end
) as deals_booked_in_month,
sum(
case when has_been_booked_within_last_6_months then 1 else 0 end
) as deals_booked_in_6_months,
sum(
case when has_been_booked_within_last_12_months then 1 else 0 end
) as deals_booked_in_12_months
from int_core__mtd_deal_lifecycle al
{% if dimension.dimension == "'by_number_of_listings'" %}
inner join
int_core__mtd_accommodation_segmentation mas
on al.id_deal = mas.id_deal
and al.date = mas.date
{% elif dimension.dimension == "'by_billing_country'" %}
inner join int_core__deal ud on al.id_deal = ud.id_deal
and ud.main_billing_country_iso_3_per_deal is not null
{% endif %}
group by 1, 2, 3
{% if not loop.last %}
union all
{% endif %}
{% endfor %}
)
-- Final aggregation of subqueries --
select
d.year,
d.month,
d.day,
d.date,
d.dimension,
d.dimension_value,
d.is_end_of_month,
d.is_current_month,
nullif(l.new_deals, 0) as new_deals,
nullif(l.never_booked_deals, 0) as never_booked_deals,
nullif(l.first_time_booked_deals, 0) as first_time_booked_deals,
nullif(l.active_deals, 0) as active_deals,
nullif(l.churning_deals, 0) as churning_deals,
nullif(l.inactive_deals, 0) as inactive_deals,
nullif(l.reactivated_deals, 0) as reactivated_deals,
nullif(l.deals_booked_in_month, 0) as deals_booked_in_month,
nullif(l.deals_booked_in_6_months, 0) as deals_booked_in_6_months,
nullif(l.deals_booked_in_12_months, 0) as deals_booked_in_12_months
from int_dates_mtd_by_dimension d
left join
deals_metric_aggregation_per_date l
on l.date = d.date
and l.dimension = d.dimension
and l.dimension_value = d.dimension_value

126
models/intermediate/core/schema.yml
View file

@ -729,42 +729,6 @@ models:
tests:
- not_null
- name: int_core__mtd_deal_metrics
description: |
This model contains the historic information regarding the deals in an aggregated manner.
It's used for the business KPIs. Data is aggregated at the last day of the month and in the
days necessary for the Month-to-Date computation of the current month.
tests:
- dbt_utils.unique_combination_of_columns:
combination_of_columns:
- date
- dimension
- dimension_value
columns:
- name: date
data_type: date
description: The date for the month-to-date deal-related metrics.
tests:
- not_null
- name: dimension
data_type: string
description: The dimension or granularity of the metrics.
tests:
- accepted_values:
values:
- global
- by_number_of_listings
- by_billing_country
- name: dimension_value
data_type: string
description: The value or segment available for the selected dimension.
tests:
- not_null
- name: int_core__mtd_guest_payments_metrics
description: |
This model contains the historic information regarding the guest payments in an aggregated manner.
@ -1495,96 +1459,6 @@ models:
data_type: integer
description: Informative field of the actual number of listings booked in the last 12 months
- name: int_core__mtd_deal_lifecycle
description: |
This model contains the historic information regarding the lifecycle of hosts, at deal level.
The information regarding the booking-related time allows for the current status of any
deal regarding its activity. This information is encapsulated in the following columns:
deal_lifecycle_state: contains one of the following states
- 01-New: Deals that have been created in the current month, without bookings
- 02-Never Booked: Deals that have been created before the current month, without bookings.
- 03-First Time Booked: Deals that have been booked for the first time in the current month.
- 04-Active: Deals that have booking activity in the past 12 months (that are not FTB nor reactivated)
- 05-Churning: Deals that are becoming inactive because of lack of bookings in the past 12 months
- 06-Inactive: Deals that have not had a booking for more than 12 months.
- 07-Reactivated: Deals that have had a booking in the current month that were inactive or churning before.
- Finally, if none of the logic applies, which should not happen, null will be set and a dbt alert will raise.
Since the states of Active, First Time Booked and Reactivated indicate certain booking activity and are
mutually exclusive, the model also provides information of the recency of the bookings by the following
booleans:
- has_been_booked_within_current_month: If a deal has had a booking created in the current month
- has_been_booked_within_last_6_months: If a deal has had a booking created in the past 6 months
- has_been_booked_within_last_12_months: If a deal has had a booking created in the past 12 months
Note that if a deal has had a booking created in a given month, all 3 columns will be true. Similarly,
if the last booking created to a deal was 5 months ago, only the column has_been_booked_in_1_month
will be false; while the other 2 will be true.
tests:
- dbt_utils.unique_combination_of_columns:
combination_of_columns:
- date
- id_deal
columns:
- name: date
data_type: date
description: The date for the month-to-date. Information is inclusive to the date displayed.
tests:
- not_null
- name: id_deal
data_type: character varying
description: Id of the deal associated to the host.
tests:
- not_null
- name: creation_date_utc
data_type: date
description: Date of when the first host associated to that deal was created.
- name: first_time_booked_date_utc
data_type: date
description: |
Date of the first booking created for a given deal. Can be null if the deal
has never had a booking associated with it.
- name: last_time_booked_date_utc
data_type: date
description: |
Date of the last booking created for a given deal. Can be null if the deal
has never had a booking associated with it. Can be the same as first_time_booked_date_utc
if the deal only had 1 booking in its history.
- name: second_to_last_time_booked_date_utc
data_type: date
description: |
Date of the second-to-last booking created for a given deal, meaning the creation
date of the booking that precedes the last one. It's relevant for the reactivation computation
on the lifecycle. Can be null if the deal has never had a booking associated with it or if
the deal only had 1 booking in its history.
- name: deal_lifecycle_state
data_type: character varying
description: |
Contains the lifecycle state of a deal. The accepted values are:
01-New, 02-Never Booked, 03-First Time Booked, 04-Active, 05-Churning, 06-Inactive,
07-Reactivated. Failing to implement the logic will result in alert.
tests:
- not_null
- name: has_been_booked_within_current_month
data_type: boolean
description: If the deal has had a booking created in the current month.
- name: has_been_booked_within_last_6_months
data_type: boolean
description: If the deal has had a booking created in the past 6 months.
- name: has_been_booked_within_last_12_months
data_type: boolean
description: If the deal has had a booking created in the past 12 months.
- name: int_core__check_in_cover_users
description:
This model contains information about hosts that offer check in cover.