Cookbook

Copy-paste starting points, drawn from the shared 54-query test corpus.

Churn (binary classification)

PREDICT NOT EXISTS(transactions.*) OVER (30 DAYS FOLLOWING)
FOR EACH customers.customer_id
WHERE EXISTS(transactions.*) OVER (90 DAYS PRECEDING)

Add RETURN PROBABILITY to get calibrated scores instead of the default output:

PREDICT NOT EXISTS(transactions.*) OVER (30 DAYS FOLLOWING)
FOR EACH customers.customer_id
WHERE EXISTS(transactions.*) OVER (90 DAYS PRECEDING)
RETURN PROBABILITY

Spend / LTV slice (regression)

PREDICT SUM(transactions.price) OVER (30 DAYS FOLLOWING) FOR EACH customers.customer_id

Recommendations (ranking)

PREDICT LIST_DISTINCT(transactions.article_id) OVER (30 DAYS FOLLOWING) RANK TOP 12
FOR EACH customers.customer_id

Daily demand, 4 weeks out (forecasting)

PREDICT SUM(usage.count) OVER (1 DAY FOLLOWING HORIZONS 28)
FOR EACH accounts.account_id

The HORIZONS 28 on the window makes this a 28-step forecast — one prediction per day.

Specific entities

PREDICT NOT EXISTS(orders.*) OVER (90 DAYS FOLLOWING) FOR users.user_id IN (42, 123)

Counterfactual

PREDICT NOT EXISTS(orders.*) OVER (90 DAYS FOLLOWING) FOR users.user_id = 42
ASSUMING users.plan = 'premium'

Status prediction (string predicate)

PREDICT LAST(loan.status) OVER (30 DAYS FOLLOWING) NOT LIKE '%DENIED' FOR EACH loan.id

Missing-attribute prediction (static target)

PREDICT articles.description IS NULL FOR EACH articles.id

Population carve-outs

PREDICT SUM(transactions.value) OVER (RANGE BETWEEN 15 DAYS FOLLOWING AND 45 DAYS FOLLOWING) > 100
FOR EACH customers.customer_id
WHERE customers.location NOT IN ('ALASKA', 'HAWAII')

As-of a fixed anchor, with quantiles

PREDICT SUM(orders.amount) OVER (RANGE BETWEEN 15 DAYS FOLLOWING AND 45 DAYS FOLLOWING)
FOR customers.customer_id IN ('C7', 'C9')
AS OF :prediction_time
RETURN QUANTILES (0.10, 0.50, 0.90)

Reusable named window

PREDICT SUM(orders.revenue) OVER w - SUM(orders.cost) OVER w
FOR EACH customers.customer_id
WINDOW w AS (30 DAYS FOLLOWING)