Python time windows

API reference

Time windows helper classes and functions

[Deprecated] WindowedDataFrame approach

Example time windowed feature

Notebook in demo repository

Imports

import os

from pyspark.sql import functions as f

Create widgets for passing in timestamp

dbutils.widgets.text("timestamp", "2020-12-12")
dbutils.widgets.text("target", "no target")

Initialise target store

%run ../init/target_store

Define time windows

time_windows=[14, 30, 90]

Load silver data

df_web_visits = spark.read.table(
    f"{os.environ['READ_ENV']}.odap_digi_sdm_l2.web_visits"
)

Add timestamp column to the dataframe by joining the target store

target_store = spark.read.table("target_store")

df = df_web_visits.join(target_store, on="customer_id").filter(
    f.col("visit_timestamp") <= f.col("timestamp")
)

Create a custom time_widowed function using visit_timestamp as your time_column