Property Management Analytics

Open Financial Dashboard Open Leasing Dashboard Open Operations Dashboard

Three connected but analytically distinct products for a conventional multifamily portfolio:

1. Financial Performance & Asset Health — where financial performance is weak and which revenue or expense categories warrant review.
2. Leasing, Occupancy & Revenue Risk — which future lease events and tenant balances require action.
3. Property Operations & Maintenance — where workload, service performance, recurring issues, or vendor patterns require intervention.

This is an independent portfolio project. Operational records are synthetic. The repository does not use or reproduce proprietary schemas or production data.

Python generates a deterministic portfolio in DuckDB, dbt applies the governed source -> stg -> base -> int -> dim/rpt transformation flow, and three Streamlit/Plotly entrypoints serve the decision workflows. The checked-in DuckDB artifact lets hosted apps start without generating data at runtime.

See architecture to learn more.

• Cutoff: June 30, 2026; history begins July 1, 2023.
• 24 properties in four U.S. metros, 3,016 units, and two property classes.
• Lease events extend through June 2027 for a twelve-month action horizon.
• Financial transactions, budgets, tenant subledger activity, work orders, status history, and vendor costs are deterministic synthetic data.
• HUD FY2026 county-level Fair Market Rent is included only as a public policy benchmark, not an asking-rent or valuation benchmark.
• Seven planted raw data-quality issues are documented in the raw-data ERD and cleaned in dbt handling.

Python 3.12 and uv are required.

make install
make build
make test

Run each product independently:

make run-financial
make run-leasing
make run-operations

The apps read data/warehouse.duckdb in read-only mode. Override it with PMA_WAREHOUSE_PATH. Cross-app deep links can be configured with PMA_FINANCIAL_APP_URL, PMA_LEASING_APP_URL, and PMA_OPERATIONS_APP_URL.

make test regenerates the fixed-seed portfolio, builds every dbt model and test, then runs Python and Streamlit application tests. make lint checks Python, SQL, model naming, dbt lineage boundaries, and deterministic SQL. CI also parses the project and generates dbt documentation from the completed build.

Financial release checks include balanced journals, AR and allocation reconciliation, and portfolio-to-property aggregation. Cross-product checks ensure contextual metrics come from the owning domain rather than being independently recalculated.

• This is a static portfolio artifact, not a live operational system.
• Forecast views are explicitly labeled scenarios or signed-lease commitments, not machine-learning predictions.
• Analysis identifies signals, arithmetic contributors, and associations; it does not claim unsupported causation.
• The multifamily model does not cover commercial, student, senior, or mixed-use lease semantics.