DocumentaçãoExperience PlatformGuia do Serviço de consulta

Recursos do engenheiro para aprendizado de máquina

Last update: Tue Jul 16 2024 00:00:00 GMT+0000 (Coordinated Universal Time)
  • Tópicos:

Criado para:

  • undefined
  • undefined

Este documento demonstra como você pode transformar dados no Adobe Experience Platform em recursos, ou variáveis, que podem ser consumidos por um modelo de aprendizado de máquina. Esse processo é conhecido como engenharia de recursos. Use o Data Distiller para calcular recursos de aprendizado de máquina em escala e compartilhar esses recursos no seu ambiente de aprendizado de máquina. Isso envolve o seguinte:

  1. Crie um modelo de consulta para definir os rótulos de destino e os recursos que deseja calcular para o seu modelo
  2. Executar a consulta e armazenar os resultados em um conjunto de dados de treinamento

Definir os dados de treinamento define-training-data

O exemplo a seguir ilustra uma consulta para derivar dados de treinamento de um conjunto de dados de Eventos de experiência para um modelo prever a propensão de um usuário para assinar um boletim informativo. Os eventos de assinatura são representados pelo tipo de evento web.formFilledOut, e outros eventos comportamentais no conjunto de dados são usados para derivar recursos de nível de perfil para prever assinaturas.

Consultar rótulos positivos e negativos query-positive-and-negative-labels

Um conjunto de dados completo para treinar um modelo de aprendizado de máquina (supervisionado) inclui uma variável de público-alvo ou rótulo que representa o resultado a ser previsto, e um conjunto de recursos ou variáveis explicativas usadas para descrever os perfis de exemplo usados para treinar o modelo.

Nesse caso, o rótulo é uma variável chamada subscriptionOccurred que é igual a 1 se o perfil do usuário tiver um evento com tipo web.formFilledOut, caso contrário é 0. A consulta a seguir retorna um conjunto de 50.000 usuários do conjunto de dados de eventos, incluindo todos os usuários com rótulos positivos (subscriptionOccurred = 1), além de um conjunto de usuários selecionados aleatoriamente com rótulos negativos para concluir o tamanho da amostra de 50.000 usuários. Isso garante que os dados de treinamento incluam exemplos positivos e negativos para o modelo aprender.

from aepp import queryservice

dd_conn = queryservice.QueryService().connection()
dd_cursor = queryservice.InteractiveQuery2(dd_conn)

query_labels = f"""
SELECT *
FROM (
    SELECT
        eventType,
        _{tenant_id}.user_id as userId,
        SUM(CASE WHEN eventType='web.formFilledOut' THEN 1 ELSE 0 END)
            OVER (PARTITION BY _{tenant_id}.user_id)
            AS "subscriptionOccurred",
        row_number() OVER (PARTITION BY _{tenant_id}.user_id ORDER BY randn()) AS random_row_number_for_user
    FROM {table_name}
)
WHERE (subscriptionOccurred = 1 AND eventType = 'web.formFilledOut') OR (subscriptionOccurred = 0 AND random_row_number_for_user = 1)
"""

df_labels = dd_cursor.query(query_labels, output="dataframe")
print(f"Number of classes: {len(df_labels)}")
df_labels.head()

Exemplo de saída

Número de classes: 50000

eventType
userId
subscriptionOccurred
random_row_number_for_user
0
directMarketing.emailClicked
01027994177972439148069092698714414382
0
1
1
directMarketing.emailOpened
01054714817856066632264746967668888198
0
1
2
web.formFilledOut
01117296890525140996735553609305695042
1
15
3
directMarketing.emailClicked
01149554820363915324573708359099551093
0
1
4
directMarketing.emailClicked
01172121447143590196349410086995740317
0
1

Agregar eventos para definir recursos para ML define-features

Com uma consulta apropriada, você pode coletar os eventos no conjunto de dados em recursos numéricos significativos que podem ser usados para treinar um modelo de propensão. Exemplos de eventos são vistos abaixo:

  • Número de emails enviados para fins de marketing e recebidos pelo usuário.
  • Parte destes emails foi aberta.
  • Parte desses emails em que o usuário selecionou o link.
  • Número de produtos que foram exibidos.
  • Número de propostas que tiveram interação.
  • Número de propostas rejeitadas.
  • Número de links selecionados.
  • Número de minutos entre dois emails consecutivos recebidos.
  • Número de minutos entre dois emails consecutivos abertos.
  • Número de minutos entre dois emails consecutivos em que o usuário realmente selecionou o link.
  • Número de minutos entre duas visualizações de produto consecutivas.
  • Número de minutos entre duas apresentações com as quais houve interação.
  • Número de minutos entre duas propostas que foram rejeitadas.
  • Número de minutos entre dois links selecionados.

A consulta a seguir agrega esses eventos:

Selecione para exibir exemplo de consulta
code language-python 
query_features = f"""
SELECT
    _{tenant_id}.user_id as userId,
    SUM(CASE WHEN eventType='directMarketing.emailSent' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "emailsReceived",
    SUM(CASE WHEN eventType='directMarketing.emailOpened' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "emailsOpened",
    SUM(CASE WHEN eventType='directMarketing.emailClicked' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "emailsClicked",
    SUM(CASE WHEN eventType='commerce.productViews' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "productsViewed",
    SUM(CASE WHEN eventType='decisioning.propositionInteract' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "propositionInteracts",
    SUM(CASE WHEN eventType='decisioning.propositionDismiss' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "propositionDismissed",
    SUM(CASE WHEN eventType='web.webinteraction.linkClicks' THEN 1 ELSE 0 END)
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "webLinkClicks" ,
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailSent', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_emailSent",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailOpened', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_emailOpened",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailClicked', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_emailClick",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'commerce.productViews', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_productView",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'decisioning.propositionInteract', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_propositionInteract",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'propositionDismiss', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_propositionDismiss",
    TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'web.webinteraction.linkClicks', 'minutes')
       OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
       AS "minutes_since_linkClick"
FROM {table_name}
"""

df_features = dd_cursor.query(query_features, output="dataframe")
df_features.head()

Exemplo de saída

userId
emailsReceived
emailsOpened
emailsClicked
productsViewed
propositionInteracts
propositionDismiss
webLinkClicks
minutes_since_emailSent
minutes_since_emailOpened
minutes_since_emailClick
minutes_since_productView
minutes_since_propositionInteract
minutes_since_propositionDismiss
minutes_since_linkClick
0
01102546977582484968046916668339306826
1
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
1
01102546977582484968046916668339306826
2
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
2
01102546977582484968046916668339306826
3
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
3
01102546977582484968046916668339306826
3
1
0
0
0
0
0
540,0
0,0
NaN
NaN
NaN
None
NaN
4
01102546977582484968046916668339306826
3
2
0
0
0
0
0
588,0
0,0
NaN
NaN
NaN
None
NaN

Combinar consultas de rótulos e recursos combine-queries

Finalmente, a consulta de rótulos e a consulta de recursos podem ser combinadas em uma única consulta que retorna um conjunto de dados de treinamento de rótulos e recursos:

Selecione para exibir exemplo de consulta
code language-python 
query_training_set = f"""
SELECT *
FROM (
    SELECT _{tenant_id}.user_id as userId,
       eventType,
       timestamp,
       SUM(CASE WHEN eventType='web.formFilledOut' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id)
           AS "subscriptionOccurred",
       SUM(CASE WHEN eventType='directMarketing.emailSent' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsReceived",
       SUM(CASE WHEN eventType='directMarketing.emailOpened' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsOpened",
       SUM(CASE WHEN eventType='directMarketing.emailClicked' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsClicked",
       SUM(CASE WHEN eventType='commerce.productViews' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "productsViewed",
       SUM(CASE WHEN eventType='decisioning.propositionInteract' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionInteracts",
       SUM(CASE WHEN eventType='decisioning.propositionDismiss' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionDismissed",
       SUM(CASE WHEN eventType='web.webinteraction.linkClicks' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "webLinkClicks" ,
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailSent', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailSent",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailOpened', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailOpened",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailClicked', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailClick",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'commerce.productViews', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_productView",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'decisioning.propositionInteract', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionInteract",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'propositionDismiss', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionDismiss",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'web.webinteraction.linkClicks', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_linkClick",
        row_number() OVER (PARTITION BY _{tenant_id}.user_id ORDER BY randn()) AS random_row_number_for_user
    FROM {table_name} LIMIT 1000
)
WHERE (subscriptionOccurred = 1 AND eventType = 'web.formFilledOut') OR (subscriptionOccurred = 0 AND random_row_number_for_user = 1)
ORDER BY timestamp
"""

df_training_set = dd_cursor.query(query_training_set, output="dataframe")
df_training_set.head()

Exemplo de saída

userId
eventType
carimbo de data e hora
subscriptionOccurred
emailsReceived
emailsOpened
emailsClicked
productsViewed
propositionInteracts
propositionDismiss
webLinkClicks
minutes_since_emailSent
minutes_since_emailOpened
minutes_since_emailClick
minutes_since_productView
minutes_since_propositionInteract
minutes_since_propositionDismiss
minutes_since_linkClick
random_row_number_for_user
0
02554909162592418347780983091131567290
directMarketing.emailSent
2023-06-17 13:44:59.086
0
2
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
1
1
01130334080340815140184601481559659945
directMarketing.emailOpened
2023-06-19 06:01:55.366
0
1
3
0
1
0
0
0
1921,0
0,0
NaN
1703,0
NaN
None
NaN
1
2
01708961660028351393477273586554010192
web.formFilledOut
2023-06-19 18:36:49.083
1
1
2
2
0
0
0
0
2365,0
26,0
1,0
NaN
NaN
None
NaN
7
3
01809182902320674899156240602124740853
directMarketing.emailSent
21/06/2023 19:17:12.535
0
1
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
1
4
03441761949943678951106193028739001197
directMarketing.emailSent
21/06/2023 21:58:29.482
0
1
0
0
0
0
0
0
0,0
NaN
NaN
NaN
NaN
None
NaN
1

Criar um modelo de consulta para calcular incrementalmente os dados de treinamento

É comum treinar novamente um modelo periodicamente, com dados de treinamento atualizados, para manter a precisão do modelo ao longo do tempo. Como prática recomendada para atualizar com eficiência seu conjunto de dados de treinamento, você pode criar um modelo a partir de sua consulta do conjunto de treinamento para calcular novos dados de treinamento de forma incremental. Isso permite calcular rótulos e recursos somente a partir de dados que foram adicionados ao conjunto de dados original do Experience Events desde que os dados de treinamento foram atualizados pela última vez e inserir os novos rótulos e recursos no conjunto de dados de treinamento existente.

Isso requer algumas modificações no query do conjunto de treinamento:

  • Adicione lógica para criar um novo conjunto de dados de treinamento se ele não existir e insira os novos rótulos e recursos no conjunto de dados de treinamento existente caso contrário. Isso requer uma série de duas versões do query de conjunto de treinamento:

    • Primeiro, usando a instrução CREATE TABLE IF NOT EXISTS {table_name} AS
    • Em seguida, usando a instrução INSERT INTO {table_name} para o caso em que o conjunto de dados de treinamento já existe

  • Adicione uma instrução SNAPSHOT BETWEEN $from_snapshot_id AND $to_snapshot_id para limitar a consulta a dados de evento que foram adicionados dentro de um intervalo especificado. O prefixo $ nas IDs de instantâneo indica que elas são variáveis que serão passadas quando o modelo de consulta for executado.

A aplicação dessas alterações resulta na seguinte consulta:

Selecione para exibir exemplo de consulta
code language-python 
ctas_table_name = "propensity_training_set"

query_training_set_template = f"""
$$ BEGIN

SET @my_table_exists = SELECT table_exists('{ctas_table_name}');

CREATE TABLE IF NOT EXISTS {ctas_table_name} AS
SELECT *
FROM (
    SELECT _{tenant_id}.user_id as userId,
       eventType,
       timestamp,
       SUM(CASE WHEN eventType='web.formFilledOut' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id)
           AS "subscriptionOccurred",
       SUM(CASE WHEN eventType='directMarketing.emailSent' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsReceived",
       SUM(CASE WHEN eventType='directMarketing.emailOpened' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsOpened",
       SUM(CASE WHEN eventType='directMarketing.emailClicked' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsClicked",
       SUM(CASE WHEN eventType='commerce.productViews' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "productsViewed",
       SUM(CASE WHEN eventType='decisioning.propositionInteract' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionInteracts",
       SUM(CASE WHEN eventType='decisioning.propositionDismiss' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionDismissed",
       SUM(CASE WHEN eventType='web.webinteraction.linkClicks' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "webLinkClicks" ,
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailSent', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailSent",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailOpened', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailOpened",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailClicked', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailClick",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'commerce.productViews', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_productView",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'decisioning.propositionInteract', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionInteract",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'propositionDismiss', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionDismiss",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'web.webinteraction.linkClicks', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_linkClick",
        row_number() OVER (PARTITION BY _{tenant_id}.user_id ORDER BY randn()) AS random_row_number_for_user
    FROM {table_name}
    SNAPSHOT BETWEEN $from_snapshot_id AND $to_snapshot_id
)
WHERE (subscriptionOccurred = 1 AND eventType = 'web.formFilledOut') OR (subscriptionOccurred = 0 AND random_row_number_for_user = 1)
ORDER BY timestamp;

INSERT INTO {ctas_table_name}
SELECT *
FROM (
    SELECT _{tenant_id}.user_id as userId,
       eventType,
       timestamp,
       SUM(CASE WHEN eventType='web.formFilledOut' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id)
           AS "subscriptionOccurred",
       SUM(CASE WHEN eventType='directMarketing.emailSent' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsReceived",
       SUM(CASE WHEN eventType='directMarketing.emailOpened' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsOpened",
       SUM(CASE WHEN eventType='directMarketing.emailClicked' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "emailsClicked",
       SUM(CASE WHEN eventType='commerce.productViews' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "productsViewed",
       SUM(CASE WHEN eventType='decisioning.propositionInteract' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionInteracts",
       SUM(CASE WHEN eventType='decisioning.propositionDismiss' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "propositionDismissed",
       SUM(CASE WHEN eventType='web.webinteraction.linkClicks' THEN 1 ELSE 0 END)
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "webLinkClicks" ,
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailSent', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailSent",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailOpened', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailOpened",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'directMarketing.emailClicked', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_emailClick",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'commerce.productViews', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_productView",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'decisioning.propositionInteract', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionInteract",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'propositionDismiss', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_propositionDismiss",
       TIME_BETWEEN_PREVIOUS_MATCH(timestamp, eventType = 'web.webinteraction.linkClicks', 'minutes')
           OVER (PARTITION BY _{tenant_id}.user_id ORDER BY timestamp ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
           AS "minutes_since_linkClick",
        row_number() OVER (PARTITION BY _{tenant_id}.user_id ORDER BY randn()) AS random_row_number_for_user
    FROM {table_name}
    SNAPSHOT BETWEEN $from_snapshot_id AND $to_snapshot_id
)
WHERE
    @my_table_exists = 't' AND
    ((subscriptionOccurred = 1 AND eventType = 'web.formFilledOut') OR (subscriptionOccurred = 0 AND random_row_number_for_user = 1))
ORDER BY timestamp;

EXCEPTION
  WHEN OTHER THEN
    SELECT 'ERROR';

END $$;
"""

Finalmente, o código a seguir salva o template de query no Data Distiller:

template_res = dd.createQueryTemplate({
  "sql": query_training_set_template,
  "queryParameters": {},
  "name": "Template for propensity training data"
})
template_id = template_res["id"]

print(f"Template for propensity training data created as ID {template_id}")

Exemplo de saída

Template for propensity training data created as ID f3d1ec6b-40c2-4d13-93b6-734c1b3c7235

Com o modelo salvo, é possível executar a consulta a qualquer momento, fazendo referência à ID do modelo e especificando a faixa de IDs de instantâneo que devem ser incluídas na consulta. A consulta a seguir recupera os instantâneos do conjunto de dados original de Eventos de experiência:

query_snapshots = f"""
SELECT snapshot_id
FROM (
    SELECT history_meta('{table_name}')
)
WHERE is_current = true OR snapshot_generation = 0
ORDER BY snapshot_generation ASC
"""

df_snapshots = dd_cursor.query(query_snapshots, output="dataframe")

O código a seguir demonstra a execução do modelo de consulta, usando o primeiro e o último instantâneos para consultar todo o conjunto de dados:

snapshot_start_id = str(df_snapshots["snapshot_id"].iloc[0])
snapshot_end_id = str(df_snapshots["snapshot_id"].iloc[1])

query_final_res = qs.postQueries(
    name=f"[CMLE][Week2] Query to generate training data created by {username}",
    templateId=template_id,
    queryParameters={
        "from_snapshot_id": snapshot_start_id,
        "to_snapshot_id": snapshot_end_id,
    },
    dbname=f"{cat_conn.sandbox}:all"
)
query_final_id = query_final_res["id"]
print(f"Query started successfully and got assigned ID {query_final_id} - it will take some time to execute")

Exemplo de saída

Query started successfully and got assigned ID c6ea5009-1315-4839-b072-089ae01e74fd - it will take some time to execute

Você pode definir a seguinte função para verificar periodicamente o status da query:

def wait_for_query_completion(query_id):
    while True:
        query_info = qs.getQuery(query_id)
        query_state = query_info["state"]
        if query_state in ["SUCCESS", "FAILED"]:
            break
        print("Query is still in progress, sleeping…")
        time.sleep(60)

    duration_secs = query_info["elapsedTime"] / 1000
    if query_state == "SUCCESS":
        print(f"Query completed successfully in {duration_secs} seconds")
    else:
        print(f"Query failed with the following errors:", file=sys.stderr)
        for error in query_info["errors"]:
            print(f"Error code {error['code']}: {error['message']}", file=sys.stderr)

wait_for_query_completion(query_final_id)

Exemplo de saída

Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query is still in progress, sleeping…
Query completed successfully in 473.8 seconds

Próximas etapas:

Ao ler este documento, você aprendeu a transformar dados no Adobe Experience Platform em recursos, ou variáveis, que podem ser consumidos por um modelo de aprendizado de máquina. A próxima etapa na criação de pipelines de recursos do Experience Platform para alimentar modelos personalizados em seu ambiente de aprendizado de máquina é exportar conjuntos de dados de recursos.

recommendation-more-help
ccf2b369-4031-483f-af63-a93b5ae5e3fb