SafeRoute

Whilst living in London, we have all felt unsafe at some point. SafeRoute is the first crime-data driven pathfinder to combat this by mitigating against danger.

Comment

Our Background:

  • We as students have all had moments where we have felt unsafe whilst living in London
  • Many Londoners, particularly women and vulnerable individuals, feel unsafe walking home after dark.
  • We personally have been harassed and been victims to theft whilst living in London

The Problem in statistics:

  • 81% of women report feeling unsafe walking home in the dark
  • 71% of people say more needs to be done to improve street safety
  • 69% We as students have all had moments where we have felt unsafe whilst living in London

Our idea:

  • Incorporate crime-data into walking directions
  • Prioritise walking routes based on safety
  • Make walking as accessible as possible

Our Solution:

  • Take in crime data from the UK Police Crime archive
  • Store and access our own Database with crime data linked to locations
  • Find the safest route based on crime data that is still viable

Building and challenges:

  • Getting crime data from data.police.uk into a database was an important consideration
  • This was mitigated via using InterSystems IRIS Data Platform
  • InterSystems IRIS vector search allows us to gain further insights into the severity of specific crime data, based on keywords
  • Our “crime context” is fully vectorized to allow for crime severity to be quantified from context from 0 to 1, for any crime.
  • Our model is trained using data obtained from vector dot product searching
  • This allows us to further avoid routes where crimes are particularly dangerous

Our path safety ranking algorithm

  • We generated multiple routes from start to endpoint, divided into coordinate pair "segments", via Google Directions API
  • We then find crimes that have occurred, within the last 3 months, along each segment of the route
  • We then compute a severity score of each crime, using our trained model, to generate a total risk score for each route and choose the safest route

Front-end

Built With

Share this project:

Updates