© 2019 Mengzhen Xiao

NYC Motor Vehicle Collisions In 2017

PROJECT BRIEF

This webpage is a breakdown of every collision in NYC in 2017 by location and injury. Each record represents a collision.

The purpose of making this map is to answer this question: How can we use a map to optimize road design? By analyzing and mapping the data, I want to find common ground in accident-prone streets locations. For urban designers, this map may also provide future directions to optimize city road design.

This webpage is a breakdown of every collision in NYC in 2017 by location and injury. Each record represents a collision.

The purpose of making this map is to answer this question: How can we use a map to optimize road design? By analyzing and mapping the data, I want to find common ground in accident-prone streets locations. For urban designers, this map may also provide future directions to optimize city road design.

DATE

April 2018 (2 weeks)

MAIN METHODS

Research

Data Analysis

Data Visualization

Front-End Development

Back-End Development

Coding Language

Mapbox gl.js

node.js

jQuery

HTML

CSS

How can we use a map to optimize road design?

The map is about the NYC motor vehicle collisions in 2017. The dataset that I used is NYPD Motor Vehicle Collisions from NYC Open Data. The purpose of making this map is to answer this question: How can we use a map to optimize road design? By analyzing and mapping the data, I want to find common ground in accident-prone streets locations. For urban designers, this map may also provide future directions to optimize city road design.

Data Analysis

The dataset I get from NYC Open Data is a huge CSV file, which has 1,048,575 rows of data from July 2013 to April 2018. Therefore, the first step is to clean the data, including formatting the data, handling missing data and weeding out nonsense information. The data I used in the final visualization is a record of all the collisions in Manhattan in 2017, which are 30,023 rows of data (30,023 collisions).

In the dataset, each record represents a collision. Since I want to know which streets are accident-prone, I have to get the data of how many collisions had occurred at each location. I’m sure there must be a simple way to get this result by running some algorithm, but for now, I did it in a stupid way that I manually went through each row and got the result. I save the result in a new CSV file.

The new CSV file to show accident-prone streets.

Building The Map

I mainly used Mapbox gl.js to build the map. The map has three parts, the overview of all collisions, serious collisions, and accident-prone streets.

Overview of All Collisions

Serious Collisions

Accident-prone Streets

The left side of the web shows the map, and on the right side is the introduction to the project. By clicking the buttons on the right, users can switch the map between the three parts. When hovering the mouse on each circle, the detail information of each collision will appear on the left corner.  For the serious collisions part and the accident-prone streets part, after clicking each circle, the Google Street View of that location will appear.

Thank You!

Read Another Case Study