← Back to portfolio
Case Study · Team Data Analysis

Addressing Crime in San Francisco

A five-person team turned a year of San Francisco's incident data into hotspots, patterns, and a funding plan — built with SQL, Python, and Tableau.

Role: Data Analyst, 5-person team (COOP Careers)  ·  Sep – Oct 2024  ·  SQL · Python · Tableau

01

The Problem

San Francisco has one of the highest property-crime rates of any major U.S. city — roughly 67 incidents per 1,000 residents, and ranked #1 nationally for property crime. Our team was tasked with one question: where, when, and what kind of crime is happening — so the city can put limited enforcement and prevention dollars where they actually move the needle.

SQL — extraction & cleaningPython — statistical analysisTableau — dashboards~150K incident records2018 SFPD open data
02

My Role

I worked as one of five analysts. I helped frame the guiding questions (where / what / when / resolution status), wrote SQL to extract and clean the raw incident feed, ran statistical analysis in Python to surface patterns, and built Tableau views to make the findings legible to non-technical decision-makers. The charts below are from our actual Tableau workbook.

03

What the data showed

Tableau map of San Francisco neighborhoods sized by incident count and colored by crime category.
Where. Incidents cluster hard in the northeast — Mission, Tenderloin, and the Financial District carry the load. Treasure Island and Seacliff are near-empty by comparison.
Line chart of monthly crime counts by neighborhood across 2018.
When. Mission leads every month; counts rise from summer into fall and spike around major holidays.
Stacked bar chart of case resolution types across police districts.
Outcome. Across every district, open/unresolved cases (purple) outnumber resolved ones — crime is reported faster than it's cleared.
04

What we recommended

01 · Prevention

Environmental design (CPTED)

Better lighting in lots and public spaces, and layouts that improve natural visibility — targeted at the larceny-from-vehicle problem.

02 · Throughput

Expanded investigation team

Crimes are committed faster than they're solved. A dedicated unit to raise the clearance rate directly attacks the open-case backlog.

03 · Recidivism

Restorative justice programs

Victim-centered healing and offender accountability to reduce repeat offending in the highest-density districts.

04 · Deterrence

Property registration

Registering high-theft property classes to make resale harder and recovery easier — lowering the payoff of opportunistic theft.

Data: SFPD Incident Reports (2018), DataSF · Census Reporter · FBI UCR. Read as direction, not precision — single-year sample.