Corona Hackathon Mock Data Generation

mocking_data.py

### IMPORTS ###
from datetime import datetime
from random import random, randint
import uuid

from faker import Faker
import numpy as np
import pandas as pd
from sqlalchemy import create_engine
from tqdm import tqdm


### CONFIG ###
DB_URL = "postgres://<USR>:<PWD>@<HOST>:5432/<DB>"
con = create_engine(DB_URL)

f = Faker("de_DE")
AGE_GROUPS = ["18-20", "21-30", "31-40", "41-50", "50+"]


### FAKER METHODS ###

def fill_employee(n=100):
    pd.DataFrame([
        {
            "id": str(uuid.uuid4()),
            "age_group": np.random.choice(AGE_GROUPS, p=[0.2, 0.2, 0.3, 0.2, .1]),
            "zip_code": f.postcode(),
            "student": random() > 0.5,
            "german_speaking": np.random.rand() > 0.5,
            "drivers_license": np.random.rand() > 0.5,
            "max_h_per_week": np.random.randint(10, 40),
            "available_from": f.date_time_this_month(before_now=False, after_now=True, tzinfo=None),
            "registration": datetime.now()
        }
        for x
        in tqdm(range(n))
    ]).to_sql("employee", con=con, if_exists="replace", index=False)

    
def fill_employer(n=100):
    pd.DataFrame([
        {
            "id": str(uuid.uuid4()),
            "name": f.company(),
            "domain": f.bs(),
            "website": f.uri()
        }
        for x
        in tqdm(range(n))
    ]).to_sql("employer", con=con, if_exists="replace", index=False)

    
def fill_job(n=500):
    
    # prep
    employer_ids = pd.read_sql_table("employer", con=con)["id"].values

    return pd.DataFrame([
        {
            "id": str(uuid.uuid4()),
            "employer_id": np.random.choice(employer_ids),
            "title": f.job(),
            "description": f.catch_phrase(),
            "zip_code": f.postcode(),
            "initial_demand_qty": np.random.randint(50, 100),
            "current_demand_qty": np.random.randint(0, 50),
            "salary_per_h": np.random.randint(12, 20),
            "start_date": f.date_time_this_month(before_now=False, after_now=True, tzinfo=None),
            "hours_per_week": np.random.randint(10, 40),
            "german_required": np.random.rand() > 0.5,
            "drivers_license_required": np.random.rand() > 0.5,
            "timestamp_posting": datetime.now(),
            "welcome_info": f.sentence(nb_words=6, variable_nb_words=True, ext_word_list=None)
        }
        for x
        in tqdm(range(n))
    ]).to_sql("job", con=con, if_exists="replace", index=False)


### EXECUTION ###
fill_employee()
fill_employer()
fill_job()