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

Merged PR 3279: Remove unused code: additive contribution to churn and its metrics

Browse source 
# Description

Simplifies the code for the models:
* int_monthly_12m_window_contribution_by_deal
* int_monthly_churn_metrics

By just removing the additive contribution approach. This also reduces the schema file information of these 2 models. I also adapted the description to clarify the state of the models.

No rush to merge this.

# 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: #22691
This commit is contained in:
Oriol Roqué Paniagua 2024-10-24 12:47:38 +00:00
parent 689ea53c11
commit 5c1f6f88f2
3 changed files with 5 additions and 132 deletions
Download patch file Download diff file Expand all files Collapse all files

36
models/intermediate/cross/int_monthly_12m_window_contribution_by_deal.sql
View file

@ -42,24 +42,6 @@ with
when ambd.preceding_months_count_by_deal < 12 then true else false
end as has_deal_been_created_less_than_12_months_ago,
-- ADDITIVE APPROACH --
-- 12 months window sum of metrics
ambd.sum_total_revenue_preceding_12_months,
ambd.sum_created_bookings_preceding_12_months,
ambd.sum_listings_booked_in_month_preceding_12_months,
-- Global amounts by sum of the 12 months window sum of metric on a given
-- date
sum(ambd.sum_total_revenue_preceding_12_months) over (
partition by ambd.date
) as global_total_revenue_preceding_12_months,
sum(ambd.sum_created_bookings_preceding_12_months) over (
partition by ambd.date
) as global_created_bookings_preceding_12_months,
sum(ambd.sum_listings_booked_in_month_preceding_12_months) over (
partition by ambd.date
) as global_listings_booked_in_month_preceding_12_months,
-- AVERAGE APPROACH --
-- 12 months window sum of metric divided by amount of active months per
-- deal
@ -101,24 +83,6 @@ select
-- Amount of active months per deal used in the computation
preceding_months_count_by_deal,
-- ADDITIVE APPROACH --
-- 12 months window sum of metrics
coalesce(
sum_total_revenue_preceding_12_months
/ nullif(global_total_revenue_preceding_12_months, 0),
0
) as total_revenue_12m_additive_contribution,
coalesce(
sum_created_bookings_preceding_12_months
/ nullif(global_created_bookings_preceding_12_months, 0),
0
) as created_bookings_12m_additive_contribution,
coalesce(
sum_listings_booked_in_month_preceding_12_months
/ nullif(global_listings_booked_in_month_preceding_12_months, 0),
0
) as listings_booked_in_month_12m_additive_contribution,
-- AVERAGE APPROACH --
-- 12 months window sum of metric divided by amount of active months per deal
coalesce(

30
models/intermediate/cross/int_monthly_churn_metrics.sql
View file

@ -19,27 +19,6 @@ with
m12wc.date,
{{ dimension.dimension }} as dimension,
{{ dimension.dimension_value }} as dimension_value,
sum(
case
when m12wc.deal_lifecycle_state in {{ churn_lifecycle_states }}
then m12wc.total_revenue_12m_additive_contribution
else 0
end
) as total_revenue_churn_additive_contribution,
sum(
case
when m12wc.deal_lifecycle_state in {{ churn_lifecycle_states }}
then m12wc.created_bookings_12m_additive_contribution
else 0
end
) as created_bookings_churn_additive_contribution,
sum(
case
when m12wc.deal_lifecycle_state in {{ churn_lifecycle_states }}
then m12wc.listings_booked_in_month_12m_additive_contribution
else 0
end
) as listings_booked_in_month_churn_additive_contribution,
sum(
case
when m12wc.deal_lifecycle_state in {{ churn_lifecycle_states }}
@ -90,15 +69,6 @@ select
d.dimension_value,
d.is_end_of_month,
d.is_current_month,
cast(
c.total_revenue_churn_additive_contribution as numeric(19, 6)
) as total_revenue_churn_additive_contribution,
cast(
c.created_bookings_churn_additive_contribution as numeric(19, 6)
) as created_bookings_churn_additive_contribution,
cast(
c.listings_booked_in_month_churn_additive_contribution as numeric(19, 6)
) as listings_booked_in_month_churn_additive_contribution,
cast(
c.total_revenue_churn_average_contribution as numeric(19, 6)
) as total_revenue_churn_average_contribution,

71
models/intermediate/cross/schema.yml
View file

@ -1249,26 +1249,13 @@ models:
- created_bookings
- listings_booked_in_month
There's 2 different contribution methods, the additive and
the average approach. The additive approach will "penalise"
accounts that have been active for less than 12 months on the
contribution amount, while the average approach "boosts" the
contribution of those accounts that have been active for less
than 12 months.
Mathematically speaking, it is encouraged to use the additive
approach ;-)
Find below the details of the computation:
- Additive approach:
Over a period of 12 months, sum the value of a given a metric
for each deal. Sum all the deals to get a global. Divide the
deal value vs. the global one.
- Average approach:
The contribution is based on an Average approach:
Over a period of 12 months, sum the value of a given a metric
for each deal, and divide it by the amount of months we're considering
for that deal. Sum all the average amounts per deals to get a global.
Divide the avg per deal value vs. the sum of avgs global one.
The average approach "boosts" the contribution of those accounts
that have been active for less than 12 months.
tests:
- dbt_utils.unique_combination_of_columns:
@ -1322,42 +1309,6 @@ models:
preceding_months_count_by_deal, and will be true on the first
year of deal activity.
- name: total_revenue_12m_additive_contribution
data_type: numeric
description: |
Share of the deal contribution on total revenue
vs. the global amount, on the preceding 12 months
with respect to date. It uses the additive approach.
It can be negative.
tests:
- not_null
- name: created_bookings_12m_additive_contribution
data_type: numeric
description: |
Share of the deal contribution on created bookings
vs. the global amount, on the preceding 12 months
with respect to date. It uses the additive approach.
tests:
- not_null
- dbt_expectations.expect_column_values_to_be_between:
min_value: 0
max_value: 1
strictly: false
- name: listings_booked_in_month_12m_additive_contribution
data_type: numeric
description: |
Share of the deal contribution on listings booked in month
vs. the global amount, on the preceding 12 months
with respect to date. It uses the additive approach.
tests:
- not_null
- dbt_expectations.expect_column_values_to_be_between:
min_value: 0
max_value: 1
strictly: false
- name: total_revenue_12m_average_contribution
data_type: numeric
description: |
@ -1407,11 +1358,11 @@ models:
This model retrieves the 12 month contribution to global metrics
by deal and aggregates it to dimension and dimension value for those
deals that are tagged as '05-Churning' in that month. Thus, it provides
a total of 6 churn related metrics, represented as ratios over the total:
a total of 3 churn related metrics, represented as ratios over the total:
- Total Revenue (in GBP)
- Created Bookings
- Listings Booked in Month
In two ways of computing the contribution, Additive and Average. For further
by using the Average contribution method. For further
information, please refer to the documentation of the model:
- int_monthly_12m_window_contribution_by_deal
@ -1450,18 +1401,6 @@ models:
tests:
- not_null
- name: total_revenue_churn_additive_contribution
data_type: numeric
description: Total Revenue churn rate (additive approach).
- name: created_bookings_churn_additive_contribution
data_type: numeric
description: Created Bookings churn rate (additive approach).
- name: listings_booked_in_month_churn_additive_contribution
data_type: numeric
description: Listings Booked in Month churn rate (additive approach).
- name: total_revenue_churn_average_contribution
data_type: numeric
description: Total Revenue churn rate (average approach).