partition_to_feature_matrix_pseudo_code.py

import featuretools as ft
import pandas as pd

def partition_to_feature_matrix(partition_num, feature_defs):
    """Calculate a feature matrix for one partition and save"""
    
    # Read in data from partition directory
    members = pd.read_csv(f's3://{partition}/members.csv')
    # ... Read in other dataframes  
    
    # Create an entityset
    es = ft.EntitySet
    es.entity_from_dataframe(id = 'customers', dataframe = members)
    # ... Add in other dataframes and relationships
    
    # Deep feature synthesis
    feature_matrix = ft.calculate_feature_matrix(entityset = es,
                                                features = feature_defs)
    # Save to S3
    feature_matrix.to_csv(f's3://{partition}/feature_matrix.csv')