/Mobility-in-US-Cities

Exploring mobility patterns in US cities

Primary LanguageJupyter Notebook

Mobility-in-US-Cities

Project exploring the mobility patterns in Chicago, Los Angeles, New York City, and San Francisco. The work leverages GPS point data scraped from cellphone/device applications. The works focuses on pattern detection and comparison across cities, network topology of mobility networks through clustering, and node-edge volume/mode prediction using machine learning.

Contributors and Authors:

  • Colin Bradley
  • Yuxuan Cui
  • Abdullah Kurkcu
  • Jianwei Li
  • Minqui Lu
  • Kaan Ozbay
  • Karan Saini

Visualizations (D3, hosted):

  • Interactive tools to visualize mobility patterns.
  • Data is cleaned and filtered data and not the entire raw data.
  • Data aggregated at different levels for various plots.
  • Please refer to github link for details.

1) City + Time wise Individual CT level data:

2) City wise Origin-Dest Puma level data:

Visualizations (Bokeh, local):

  • Interactive tools for EDA to check data quality.
  • Data is sponsor provided raw data.
  • Data aggregated at CT level for all plots.
  • Please refer to python notebooks with same name as plots for details.

1) Mode Time Map

2) Bokeh Heat Map

3) Bokeh Arc Map

Visualizations (NetworkX, local):

  • View of the Origin-Dest data in the form of a network.
  • Nodes represent centroids of PUMA level geometry.
  • Edge weights represent aggregated GPS pings.