STRAVA Spatial Data and Processes: Utilizing GIS and spatial tools to generate new data products from raw GPS cycling tracks on Strava Introduction and Thank you.
Strava’s Big Spatial Data 125+ million activities 300+ billion GPS points 60% international Strava, Inc. 2014
Strava Labs Projects Strava, Inc. 2014
How did the Geo Pod come to be? Geo Reporting Bad GPS data Segment corrections www.strava.com/routes Geo Team was created With over 400 billion Gps points we have global data coverage. Cutting edge routing Initial work with ODOT, NYU This data needs to be in the hands of planners. To many regions around the world have zero idea of where any one is riding in there region. Strava, Inc. 2014
Strava Metro Quick Timeline Data requests in early 2012 Planning, Research, Big Data Raw data No raw data sold or provided to groups Strava Metro was born and first release to Oregon DOT on February 12, 2014 Strava, Inc. 2014
Huge Amounts of Data Needed: What can GEO/GIS fill in? Restaurant Dev zones Where do cyclist spend Money Difference between men and women cyclists? Bike peer-networks Bike new or used % restaurant sales by bike How to gather feedback on design elements Perceived dangerous or challenging areas What is the mode share by street segments Where do people ride vs where do the want to ride bike ownership rates by location and demographics Where are bikes stored Bike purchasing after bike share opens Heath insurance for those who work on a bike Bike theft data Crash data based on cause/mode Bikes on transit Perception of bikes and bicyclists Locations of cycling habitat fragmentation Perception of safety on various faculties Bike tire and accessory sales data What is the rate of adoption Loner bike for visitors Patterns of usage that suggest better, safer ways of designing streets Bike share routes vs. bike network Where are bikes parked Health of riders over time Locations of near-misses and close calls Cycling demo- who is riding Cycling lanes not used Lidar of street surface Traffic counters % of journeys on bike lanes Free indoor parking? Structure, qualitive rating of route or route segment What makes new bike commuters starts? How many people starting or re-starting riding because of bike share Meeting with planners, researchers and cyclists has generated a huge gap in data. Strava can fill a lot of this! Strava, Inc. 2014
Who makes up San Francisco Strava? Who is Strava? Strava has grown into the leading online cycling and running community Our members cross all types of demographics Urban area uploads are around 40% commute Who makes up San Francisco Strava? Athlete ID Count: 23,714 Activity Count: 394,883 Average Distance: 28,904 m Median Distance: 19,409 m Average Time: 6,442 sec Median Time: 4,188 sec Commute Counts: 183,155 Male Count: 18,511 Male Count Under 25: 670 Male Count 25 - 34: 5,826 Male Count 35 - 44: 5,145 Male Count 45 - 54: 3,444 Male Count 55 - 64: 1,129 Male Count 65 - 74: 243 Male Count 75 - 84: 22 Male Count 85 - 94: 19 Male Count No Bday: 2,004 Female Count: 3,705 Female Count Under 25: 123 Female Count 25 - 34: 1,501 Female Count 35 - 44: 841 Female Count 45 - 54: 448 Female Count 55 - 64: 166 Female Count 65 - 74: 30 Female Count 75 - 84: 1 Female Count 85 - 94: 1 Female Count No Bday: 591 Blank Gender Count: 1,498 Strava is not just cat1 male racers. FB like funnel growth Cyclists like to upload. Strava has a great permanent social home feel. No ride is to small and no ride is to large. We view cyclists the same. Strava, Inc. 2014
Locating Commutes A key question is do we have commute data and if so how do we find it. We have spent a long time working through this. There are 3 key was we do this: Native Strava flag from the website (there was once a black jersey for leading commuters, carbon offset), Fuzzy name matching from cycling titles, GIS Point 2 point /w distance and time constants. Strava, Inc. 2014
Strava Metro is a Custom Built Product Vector GIS polyline Layers Basemap is user dependent (OSM, TomTom, City Provided) AM/PM commute times On and Off Season Polygon tables: Census Blocks, zip codes, wards or custom City Bike & CycleTracks GPS Tracks Integration Strava is dynamic and is built to have seamless integration into a planning department. User defined fields allow for a deep understanding of the data provided. Strava, Inc. 2014
What does strava do. Blending GIS and Programming What does strava do? Blending GIS and Programming. Taking noise and huge amounts of data and making them useable and ideal for deep analysis. Polylines do not show direction, time. Points are a cluster that can be challenging to work through. Strava’s custom format allows for quick data extraction and clean views of the data. Protecting the users privacy and not selling their raw data is very important to us. Strava, Inc. 2014
Core Strava Metro Files Strava Metro is really a suite of products with the Minute data being the key data to the group. These tables can be very large and often require a Postgres like database to use. We provide a set of rolled up data with the Minute data to allow for quick views od the data. These views are agreed upon between Strava and each group. Strava, Inc. 2014
Minute View Strava, Inc. 2014
Rolled Up View The wiggle Strava, Inc. 2014
Filtering by Commutes Strava, Inc. 2014
Origin/Destination Matrix The OD was the first logical new data feed for the Strava Metro suite. The data in this format shows great functionality of zonal transportation by bike. When paired with the minute routing data if then shows the key paths used between each zone. Strava, Inc. 2014
Spatial Movement by Zones Strava, Inc. 2014
Node Wait Times To wait or not to wait. How long will a cyclist wait? What the busiest intersections in the city? Do cyclists stay away from them? These are all the questions that the Nodes product will answer. Strava, Inc. 2014
Intersection Based Data Strava, Inc. 2014
Strava Metro: The Big Picture Strava, Inc. 2014
Strava Metro Infrastructure Change Adoption Using Strava provides quick data back that can be used to evaluate the impacts and effectiveness of changes to cycling lanes or area. Strava, Inc. 2014
Net Gain & Loss Green + 100 bike trips and blue – 100 bike trips May to July. Strava, Inc. 2014
Weekend vs Weekday Use Strava Metro can be used to see quick trends or at the same time the data can be blended to show street by street differences. Strava, Inc. 2014
Peak Riding Times and Days Within 2 minutes of using the data you can pull out AM vs PM cycling peaks and seasonality trends. This can be then broken down even further to show trends at 8am for the year for one piece of road. Strava, Inc. 2014
Strava Correlation: Oregon Sample Correlation to cycling counters is key for many groups. This is because it let’s you start to visualize the data on every road from the % match at the counters. Strava market saturation varies by region but we have found that in mature areas its between 2 – 12 %. Strava, Inc. 2014
Web Interface: OD and Streets Example Many groups do not have the ability to use a GIS platform so we will start to offer a Strava Metro online experience that will provide insight for everyone into cycling. This is designed primarily for advocacy groups and MTB network managers. Strava, Inc. 2014
Missing Cycling Corridors Cyclists often take the less beaten path. This involves cut through, off road paths, walking with bike up stairs and so on . The missing corridors product will start to highlight these paths so that planners can start to understand why its happening and create a safer solution. Strava, Inc. 2014
Final Metro Points Strava is receiving over 3 million uploads a week and has doubled every year Strava is free to download/use, ad free and available on Iphone and Android phones Strava’s apps are translated in 13 languages and the website in 5 Strava GEO is pushing the edge of providing critical cycling data to planning groups Strava, Inc. 2014
Animation Time http://vimeo.com/102942524 https://www.youtube.com/watch?v=QZ5DuQTUPqk&list=UUdO8l6B6yeRDcsP1snssTPw Strava, Inc. 2014
Strava, Inc. 2014