Types

Base type to use with autofeasts.

This will init a dataclass with all neccessary information to build features with autofeats. This class also creates a join between table and public to create a time window used in features creation.

The join works creating a on clause with the id (for example a customer_id) and dates columns.

• Example:

  If you have a data_ref equals to “2022-01-06” and defines subtract_in_end parameter equals to 5 you will select all transactions from “2022-01-01” until “2022-01-05”.

class autofeats.types.Dataset(table: pyspark.sql.dataframe.DataFrame, primary_key_col: str, table_join_key_col: str, table_join_date_col: str, numerical_cols: list, categorical_cols: list, public: pyspark.sql.dataframe.DataFrame, public_join_key_col: str, public_join_date_col: str, subtract_in_start: int = 0, subtract_in_end: int = 90, time_unit: str = 'day')

Bases: object

categorical_cols: list

List of categorical columns in the dataframe

numerical_cols: list

List of numerical columns in the dataframe

primary_key_col: str

Primary key of the table

public: DataFrame

Public dataframe

public_join_date_col: str

Public date column to use in the join

public_join_key_col: str

Public ID to use in the join

subtract_in_end: int = 90

Time window lenght

subtract_in_start: int = 0

Days before the reference date

table: DataFrame

Dataframe to create features

table_join_date_col: str

Date column used to do the join with public to create a time window

table_join_key_col: str

Column used to do the join with public

time_unit: str = 'day'

Unit used to make time window