This is the reference doc. See it running on real generated data, or build it yourself in the Studio.
On this page
Generate Real Estate Synthetic Data in Python
Real estate data has patterns that random generation misses: home prices follow a lognormal distribution with a heavy right tail (the $1M+ tier), days-on-market is lognormal with a median around 23 days, and agent ratings cluster toward 4–5 stars due to platform selection effects. Misata generates a three-table real estate dataset, agents, properties, and transactions, where these distributions are built in and FK relationships are always valid.
Python
import misata
tables = misata.generate("A real estate agency with 500 properties and 50 agents", rows=500, seed=42)
print(list(tables.keys())) # ['agents', 'properties', 'transactions']
print(tables["properties"][["price", "bedrooms", "status"]].describe())What Misata generates#
Three tables: agents (who manage listings), properties (listed inventory), and transactions (closed sales). Every property references a valid agent; every transaction references a valid property.
Tables and columns#
| Table | Key columns |
|---|---|
agents | agent_id, name, email, agency, rating, listings_sold, years_experience |
properties | property_id, agent_id, address, city, state, price, bedrooms, bathrooms, sqft, listing_date, status |
transactions | transaction_id, property_id, buyer_name, sale_price, close_date, commission_rate, days_on_market |
Realistic distributions#
- Home prices lognormal with US median ~$410k, realistic heavy right tail for luxury properties
- Days on market lognormal with median ~23 days, fast movers and long-stale listings both present
- Agent ratings beta-distributed toward 4–5 stars, reflecting real platform rating dynamics
- ~60% of listings close: remainder are active, expired, or withdrawn
sqftandpriceare correlated, larger properties cost more
Quick start#
Python
import misata
tables = misata.generate("A real estate agency with 500 properties and 50 agents", rows=500, seed=42)
# Price distribution
print(tables["properties"]["price"].describe())
# Listing status breakdown
print(tables["properties"]["status"].value_counts())
# Commission revenue
transactions = tables["transactions"]
transactions["commission"] = transactions["sale_price"] * transactions["commission_rate"]
print(f"Total commission revenue: ${transactions['commission'].sum():,.0f}")Common use cases#
- MLS / property search platform development: seed a test database with listings across price tiers and bedroom counts for search, filter, and sort testing
- AVMs (automated valuation models): generate training data with realistic price, sqft, bedrooms, and location features for regression model development
- Agent performance analytics: build conversion rate, average days-on-market, and commission dashboards on realistic agent histories
- CRM for real estate agents: test lead management and listing pipeline workflows with realistic property and transaction data
- Proptech demo environments: replace real MLS data exports with synthetic equivalents for vendor demos and investor presentations
- Market report generation: validate report templates and calculations against a full year of synthetic transaction data
Advanced: market cycle narrative#
Python
tables = misata.generate(
"Real estate market with rising prices through mid-year, rate hike slowdown in Q3, "
"slight recovery in Q4",
rows=1000,
seed=42,
)Advanced: locale-aware generation#
Python
# UK property market — GBP pricing, British address format, UK cities
tables = misata.generate("UK estate agency with 300 properties in London and Manchester", rows=300)
# Australian market — AUD pricing, Australian cities, state codes
tables = misata.generate("Australian real estate agency with 400 properties", rows=400)Advanced: quality-guaranteed generation#
Python
tables = misata.generate(
"Real estate market with 500 listings",
min_quality_score=85,
smart_correlations=True, # auto-adds sqft↔price, bedrooms↔price correlations
rows=500,
seed=42,
)