lag_timesteps

Description

For forecasting problems, leverage the auto-regressive labels as inputs. This parameter controls the number of previous values should be considered as auto-regressive labels.

• The unit of the forecasting step is the same as the unit in the aggregation (default days).

Supported Task Types

• Forecasting, Regression

Example

For the following query, a value of lag_timesteps=3 will consider the last 3 weeks' data points where each of the 3 values are aggregated on 7-day intervals.

PREDICT SUM(TRANSACTIONS.AMOUNT, 0, 7, days)
FOR EACH CUSTOMERS.CUSTOMER_ID

🚧

NOTE: lag_timesteps is currently experimental.