+ All Categories
Home > Documents > Safest Route Prediction in Urban Areas Final Presentation

Safest Route Prediction in Urban Areas Final Presentation

Date post: 09-Jul-2016
Category:
Upload: gabriel-thomas
View: 5 times
Download: 0 times
Share this document with a friend
Description:
This is our Advanced big data project, where we try to plot a safe route based on crime history.Gaussian Mixture model was chosen to model the distribution of crime.The Naive assumption that 24 hour shops are safe was used to plot safe waypoints based on the value of the gaussian in those locations and their distance from previous waypoints.
13
© 2015 CY Lin, Columbia University E6895 Advanced Big Data Analytics Final Project Presentation 1 E6895 Advanced Big Data Analytics: Safest Route Prediction in New York City May 12, 2016 Team Members: Gabriel Thomas (gtm2122), Anubha Bhargava (ab3955)
Transcript
Page 1: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

1

E6895 Advanced Big Data Analytics:

Safest Route Prediction in New York City

May 12, 2016

Team Members: Gabriel Thomas (gtm2122), Anubha

Bhargava (ab3955)

Page 2: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Motivation

● Develop a useful, problem solving tool that displays the safest walking

route in a city

● The application will be designed for those unfamiliar with a city or

uncomfortable walking at night.

The application will…

● Display the safest route on a map interface

● Provide walking directions to the user

Page 3: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Datasets, Software Languages and APIs Used

Datasets: NYPD Major Felony Incidents Crime Dataset, spotcrime.com

Software Languages and packages: Python, JavaScript, HTML, Spark

APIs: Yelp, imaplib, GoogleMaps, MapQuest

Page 4: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Front-End Interface

Page 5: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Visualization Map

Page 6: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Algorithm

1. Get 24 hour shops from Yelp API

2. Load historical crime datasets and Spotcrime.com data

3. Get circle around the origin and destination

4. Get all the coordinates for 24 hour shops within this area

5. Use Gaussian Mixture Modelling to fit the crime locations

6. Use the mixed Gaussian Multivariate Distribution on the locations of the

24 hour shops to check safety

7. Use these locations as waypoints to plot the route in Google Maps and

provide walking directions

8. Plot the crime data on using a visualization plot on Google Maps

Page 7: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Safest Walking Route Prediction

Page 8: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Safest Walking Route Prediction

Page 9: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Safest Walking Route Prediction

Google Maps Original Route:

Page 10: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Safest Walking Route Prediction

Page 11: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Safest Walking Route Prediction

Google Maps Original Route:

Page 12: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Opportunities for Future Development

● Instead of only using 24 hour shops, using additional waypoints would

aid in determining safety.

○ A more precise technique would be determining the safety at the

coordinates of each leg.

● The current algorithm holds each crime at the same weight, but a

more robust algorithm would hold more dangerous crimes with more

weight.

● Each crime is currently modeled as a independent and identically

distributed Gaussian mixture model.

○ It would be useful to experiment with different mixture models.

Page 13: Safest Route Prediction in Urban Areas Final Presentation

© 2015 CY Lin, Columbia UniversityE6895 Advanced Big Data Analytics – Final Project

Presentation

Questions?


Recommended