p_03 · Public Safety · AI-built report
Car Theft Analytics + AI
New Zealand vehicle theft analysis on 4,553 records. I designed the star schema and 28 DAX measures. Claude Code built the 5-page report and theme under my direction.
- Power BI
- DAX
- Power Query
- Claude Code
01 / Problem
How many vehicles were stolen in New Zealand, where, which vehicles, and when? And can AI build the report layer while a BI developer owns the model?
02 / Data
- Three CSVs from NZ Police stolen vehicle records: 4,553 thefts, 138 makes, 16 regions with population and density.
- Period: 7 Oct 2021 to 6 Apr 2022.
- Cleaned in Power Query: unknown makes mapped, blanks standardised, population text converted, dates parsed.
03 / Approach
- Me: star schema with 1 fact and 4 dimensions, 28 DAX measures in 6 display folders (time intelligence, rates per 100K, rankings), optimisation rules.
- Claude Code: 5 storytelling pages, custom theme, page navigation, then validated each page from screenshots.
- I reviewed every round and pushed back until navigation, titles and colour worked.
stage_1
3 CSV files
NZ Police records
stage_2
Power Query
Cleaning
stage_3
Star schema
1 fact · 4 dims
stage_4
28 DAX measures
6 folders
stage_5
5-page report
Built by Claude Code
04 / Live dashboard
05 / Business impact
- Monthly thefts more than doubled, from 464 (Oct 2021) to 1,053 (Mar 2022).
- Auckland has 36% of thefts, but Gisborne has the highest rate at 338 per 100K.
- Only 4.2% of stolen vehicles are luxury makes. Toyota, stationwagons and silver cars lead.
06 / Tools
- Power BI Desktop
- PBIP / TMDL
- Power Query (M)
- DAX
- Claude Code