slides-eurostat16



slides-eurostat16

0 0


slides-eurostat16

Slides for CartoDB Eurostat talk

On Github jsanz / slides-eurostat16

Making visible the invisible

the L Train closure

Jorge Sanz · jorge@cartodb.com · @xurxosanz

Eurostat 2016 Visualization Workshop

Agenda

  • Problem definition
  • Current status
  • Proposed solution
  • Technology

About us

  • CartoDB: location intelligence SaaS
  • Since 2011
  • ~100 employees
  • Offices in Madrid, NYC and Tartu
  • ~200K users

Problem definition

  • After Sandy hurricane, the NY Canarsie tunnel was affected
  • Authorities (MTA) are planning to shut down the L train line
  • This situation can last years
  • How this line shutdown will affect to New Yorkers?

Overview the L Train

Overview the L Train

  • Major connection between Manhattan an Brooklyn
  • ~27.000 passenger per workday (2014 data)
  • With its modern signaling, L line can have more trains

Current status

Data available

Analysis

Objective: study the relation between people from Brooklyn commuting to Manhattan and the L line.

  • Census blocks as base geographic information
  • Relation with L entrances using walking route and travel time

Walking paths

Map of number of Brooklyn walkers using L train within a 30 minutes range

SE and NW areas of Brooklyn are not covered by any other subway. We limited to 30 minutes ride and Census data doesn't reflect commuters taking a bus to arrive to the subway entrance.

Commuting time

  • ~500K people from Brooklyn work at Manhattan
  • ~200K L Train commuters per day
  • ~114K use the L train to work at Manhattan

People try to live close to their jobs

Scenario

closing the Canarsie tunnel

Affected population

L is the closest line for ~19K low income households

Affected population

Affected population

  • Most affected areas would be Brownsville and East New York
  • More wealthy population can look forother transport methods (uber, lyft)
  • MTA suggested bus shuttles

Proposed solution

Bus shuttles to substitute the Canarsie tunnel

  • Williamsburg to Lower East Side
  • ~20 minutes ride
  • Is it worth?

Bus shuttle: best choice?

Bus shuttle: best choice?

Shuttle is the best choice for ~75.000 riders per day

  • 1154 bus rides (65 pax/bus)
  • One bus every 90 seconds
  • This almost means NYC needs a new bridge!!

Increasing bus time

  • Shuttle are only interesting if they take less than 29 minutes
  • Then all commuters would better use M line

Technology

CartoDB as an analysis platform

  • Postgres + PostGIS
  • Mapzen Valhalla routing exposed as SQL functions
  • Custom python functions embedded

CartoDB Data Services API

Expose third party services through SQL:

  • Geocoding: Mapzen, HERE, Google
  • Routing: Mapzen
  • Trade Areas: HERE

Example: Geocoding + Routing

SELECT
  st_transform(shape, 3857) as the_geom_webmercator
FROM
cdb_route_point_to_point(
  cdb_geocode_street_point('Calle Moreno Nieto, 2, 28005 Madrid'),
  cdb_geocode_street_point('fernandez de la hoz 35, madrid'),
  'car',
  ARRAY['mode_type=shortest']::text[]
)

Technology: geocoding + routing

Looking forward

  • A beter UI
  • Reference data
  • More analysis functions, beyond PostGIS
  • Better LDS

For everyone, everything at https://github.com/CartoDB/

Thanks!

Making visible the invisible: the L Train closure

Jorge Sanz · jorge@cartodb.com · @xurxosanz

Credits

1.1
Making visible the invisible the L Train closure