File size: 1,337 Bytes
8d6dab8
6babaf2
 
 
 
8d6dab8
 
6babaf2
 
8d6dab8
 
6babaf2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
---
title: redrawn
emoji: 🗺️
colorFrom: blue
colorTo: gray
sdk: static
pinned: false
license: mit
short_description: Redraw the district lines and the answer changes
---

# redrawn

**Does the map change the answer?** 1.9 million New York City crashes, one
question — do neighbourhoods where speeding is cited more often have more
injurious crashes? — and a different answer from every map of the city.

Between individual crashes the correlation is **+0.055**. Across census tracts
it is +0.262, across police precincts +0.469, across community districts +0.551,
across boroughs **+0.709**. Contiguous equal-size districts drawn deliberately
span **+0.07 to +0.83**. Not one crash record changes; only the boundaries move.

This page re-runs the analysis in your browser. It holds the per-tract crash
counts and the tract adjacency graph and recomputes the zone rates and the
correlation on every interaction, so nothing you see is a stored answer — and
the "draw a new map" button really does build a fresh contiguous partition of
2,310 census tracts.

Data: [NYPD Motor Vehicle Collisions](https://data.cityofnewyork.us/Public-Safety/Motor-Vehicle-Collisions-Crashes/h9gi-nx95)
and Department of City Planning boundaries, via NYC Open Data.

Method, controls and the full write-up: <https://github.com/UsmarHaider/redrawn>