VedAustin

import logging
import timeit
from dataclasses import dataclass
from functools import partial
from statistics import median

logging.basicConfig(level=logging.INFO)


@dataclass(slots=True)

VedAustin

from pyspark.sql import functions as F
from pyspark.sql import Window

# Read data
user_guid_email = spark.read.json("/mnt/public-blobs/attribution-modelling/data2/id-maps/id-map-email.json")
user_guid_paid_search = spark.read.json("/mnt/public-blobs/attribution-modelling/data2/id-maps/id-map-paid-search.json")
user_guid_social = spark.read.json("/mnt/public-blobs/attribution-modelling/data2/id-maps/id-map-social.json")

guid_event_email = spark.read.parquet("/mnt/public-blobs/attribution-modelling/data2/events-email")
guid_event_paid_search = spark.read.parquet("/mnt/public-blobs/attribution-modelling/data2/events-paid-search")