Nowadays many communities in the United States and Canada provide a special telephone number "3-1-1". It works just like 9-1-1, but for request of non-emergency municipal services, such as: a broken streetlamp, presence of rats in a neighbourhood, bulk trash pick up. Inhabitants can submit their requests either by phone or by using a web platform.
In NYC the service has been introduced in 2003 and became the largest 3-1-1 in operation to date. All those requests are made publicly available with a lot of information regarding each ticket, such as: time of submission, time of closing, neighborhood, geolocation, etc. Through analysis of this data we aim to understand what are the main problems in NYC. A success in providing meaningful insights could be used as a proof of concept to convince more cities to set this service up and ease the communication between a city and it's inhabitants.
Based on our conclusions we will provide advice to the New York public service providers.
- Identify recurring patterns and evolution of the complaints relating to rats.
- Rank neighborhoods based the number of complaints and delay in their resolution.
- Identify major problems in boroughs tackled by NYPD through 311.
- Identify correlation between certain events, such as earthquakes or hurricanes, and the complaints that follow them.
- NYC 311 Service Requests from 2010 to Present
All the 311 requests from 2010 up till now, updated daily. Currently about 22M entries, publicly available. The data is stored in csv format and contains attributes like: Complaint Type, Created Date, Closed Date, Due Date, Incident Address, etc. - NYC Population By Neighborhood Tabulation Areas
Self explanatory. In csv format. - NYC Population By Borough
Self explanatory. In csv format.
Boundaries of areas in NYC in GeoJSON format:
- Neighborhood Tabulation Areas
- Police Precincts
- School Districts
- DSNY Districts
- Fire Battalions
- The data can be accessed from data.cityofnewyork.us
- The 311 dataset contains location based on GPS coordinates while the population data is based on the Neighborhood Tabulation Areas, so we will have to map the requests to regions using the Neighborhood Tabulation Areas dataset.
- Analyse complaint types and dates to identify short-term and long-term trends for Neighborhood Tabulation Areas, seasonalities, and separate the changes within neighborhoods from city-wide trends.
- Compile a clean dataset of the external events related to neighborhoods, and attempt to provide insights on the changes observed by relating event timestamps to our trends.
- Provide a insights about neighborhoods using relevant metrics.
- Artur Szalata: Parts of the data-story: from 311 plot of the calls over time till the word-cloud (inclusive on both sides), analysis on departments and boroughs, conclusion. Cleaning the dataset. Analysis of the departments of agencies in the appendix.
- Louis Landelle: Complaint type class classification; analysis, normalization of complaint type class pest; the functions to generate time geojson features and folium timelapses.
- Julien Heitmann: Worked on the analysis of the noise complaints, and the NTA ranking. Wrote a script to process the dataset using Python's multiprocessing library.
- Olivier Cloux: Worked on analyzing and displaying seasonality pattern in the dataset. Also, created most of the structure of the website, and made it responsive, modern, and concise. Design of poster and presentation. All of us will work on the poster and presentation.