jdwyah

create or replace function
 public.significance(control_size integer,
              control_conversion integer,
              experiment_size integer,
              experiment_conversion integer)
 returns float
 stable as $$
   from scipy.stats import norm

   def standard_error(sample_size, successes):

jdwyah

     daily AS (

        SELECT
          experiment_arm.day,
          experiment_arm.experiment_test,
          experiment_arm.cumulative_exposures :: INT      experiment_cumulative_exposures,
          experiment_arm.cumulative_order_placed_1 :: INT experiment_cumulative_conversions,

          control_arm.cumulative_exposures :: INT         control_cumulative_exposures,
          control_arm.cumulative_order_placed_1 :: INT    control_cumulative_conversions

jdwyah

      cumulative AS (
        SELECT
          DAY,
          experiment_test,
          experiment_result,
          exposures,
          sum(exposures)
          OVER (
            PARTITION BY experiment_test, experiment_result
            ORDER BY DAY

jdwyah

      results AS (
        SELECT
          event,
          experiment_test,
          experiment_result,
          day,
          count(DISTINCT first_exposures.tracking_id)            exposures,
          count(DISTINCT conversions_order_placed_1.tracking_id) order_placed_1,
          count(DISTINCT conversions_order_placed_1.tracking_id) /
          count(DISTINCT first_exposures.tracking_id) :: FLOAT   conversion_rate

jdwyah

 WITH first_exposures AS (
      SELECT
        tracking_id,
        created_at,
        event,
        experiment_test,
        experiment_result,
        date_trunc('day', created_at) AS day
      FROM warehouse.events
      WHERE category = 'experiment'

jdwyah

CREATE TABLE events (
  event_guid           CHARACTER VARYING(128),
  tracking_id          CHARACTER VARYING(128),
  category             CHARACTER VARYING(128) encode TEXT255,
  sub_category         CHARACTER VARYING(128),
  event                CHARACTER VARYING(128) encode TEXT255,
  event_name           CHARACTER VARYING(128) encode TEXT255,  
  value_decimal        NUMERIC(38, 6),
  value_string         CHARACTER VARYING(255),
  created_at           TIMESTAMP WITHOUT TIME ZONE,

jdwyah

WITH monthly_usage AS (
    SELECT
      user_id,
      date_part('month', age(created_at, '1970-01-01')) +
      12 * date_part('year', age(created_at, '1970-01-01')) AS time_period
    FROM orders
    WHERE order_state = 'completed'
    GROUP BY 1, 2
    ORDER BY 1, 2)

jdwyah

SELECT
  date_trunc('day', page.timestamp)
  count(distinct page.anonymous_id),
  count(distinct conversions.user_id)
FROM pages page
  JOIN identifies ids
  LEFT JOIN conversions ON conversions.user_id = identifies.user_id
WHERE
  ids.anonymous_id = page.anonymous_id
  AND completes.user_id = ids.user_id

jdwyah

create table events (
user_id int
date date_time
event text
)

jdwyah

calculated as (select time_period,
  case when lag is null then 'NEW'
     when lag_size = 1 then 'ACTIVE'
     when lag_size > 1 then 'RETURN'
  end as this_month_value,
 
  case when (lead_size > 1 OR lead_size IS NULL) then 'CHURN'
     else NULL
  end as next_month_churn,