Pro Bono Project · Loan Distribution Analysis · Spatial Analysis & Data Visualization & Tableau

Loan Distribution & Fairness/Equity Analysis: A state-level total can't tell you where lending actually concentrates — or for whom.

Pro bono project delivered through Byteforza for a US-based founder · home mortgage lending across US banks · fairness/equity analysis · spatial (geospatial) analysis · interactive data visualization · drill-down from state → county → census tract.

A fair-lending review needs to know whether home mortgage approvals cluster unfairly by geography or race — and a single portfolio-wide number can't answer that. I built a Tableau workbook that drills from state, to county, to individual census tract, filterable by race, bank, and year at every level — so a pattern that's invisible in an aggregate becomes a specific, visible cluster of tracts on a map.

Open the live workbook on Tableau Public → My Tableau Public profile →

Who Fair-lending compliance analysts Need to check whether lending concentrates unfairly by geography or race — not just read a portfolio-wide total.

What Drill from state to county to census tract Filterable by race, bank, and year at every level — turns "where's the disparity" into a clickable map, not a spreadsheet pivot.

Why Aggregate numbers hide exactly what compliance needs to catch A state total can look perfectly balanced while individual tracts inside it don't — tract-level granularity is what makes that visible.

Loan Fairness Distribution workbook — drilling from state view to county view to census-tract view, filtered by race and bank — The workbook, in motion — state view → county view → tract view, with race/bank/year filters applied at each level.

Two things a state-level total can't show

One number hides where it's concentrated

A state's total loan volume can look unremarkable while the county or tract underneath it tells a completely different story — the aggregate is exactly the level at which a disparity disappears.

No way to check for racial disparity without filtering

Without a race filter available at every geographic level, a reviewer can't isolate whether loan amounts or approval patterns differ by race within the same county or tract.

What it does

Three geographic levels, one workbookState → county → census tract, each its own map view, not three separate reports.
Filterable by race, bank, and yearThe same three filters apply at every level, so a pattern can be traced from a state down to the specific tracts driving it.
Color-coded by loan amountLeast-to-highest color scale at every level, so concentration is visible at a glance before any filter is even applied.
Built on US Census geographyCounty and tract boundaries come from Census TIGER/Line shapefiles, joined to loan records via FIPS codes — not hand-drawn regions.

Impact in numbers, and the baseline

Delivered to the founder as an analysis workbook, not a deployed compliance tool with a measured before/after, so no business-outcome baseline exists. The numbers below describe the analytical capability, not a measured result.

3 levels

State → county → census tract

one workbook, not three separate reports

3 filters

Race, bank, and year

applied consistently at every geographic level

$249M–$580M

State-level loan volume range shown

example state view · least-to-highest color scale