Azavea - Do Good Data 2015

A presentation at Do Good Data 2015 by Tyler Dahlberg and Daniel McGlone on behalf of Azavea. About bicyle theft, pedestrian-auto crashes, case studies, and GIS analysis and webmapping.

Transcript

1. Breaking the Cycle: When Spatial Analysis & Transit Data Collide bit.ly/DGDGIS
2. bit.ly/DGDGIS 1. Case Study - Bicycle Coalition 2. Case Study - PlanPhilly 3. Demo - QGIS 4. Demo - CartoDB 5. Recommended tools
3. Civic Software, Data Analytics & Research
4. B Corporation • Civic / Social Apps • Pro Bono Program • Donate share of profits Research-Driven • 10% Academic Research Program • Academic Collaborators • Open Source • Open Data
5. Dan McGlone • Intrepid Explorer • Suave mapadelic • Cycling Epicurean
6. Tyler Dahlberg • Former Iowan • Adventurous Mapper • Fearless Hamburger-eater
7. More at Summerofmaps.com • $5,000 Fellowship for Student GIS Analysts • Work on pro-bono projects for nonprofits • Mentored by Azavea Staff
10. Stolen: July 2010
12. Stolen: May 2013
13. How many stolen in Philly from 1/2007 - 12/2012?
14. 10,747
15. $3,900,000
16. What about Chicago? • http://data.cityofchicago.org • Only 4,652 in twice the length of time? • May be due to CPD’s insistence on serial #s
17. Simplify complex patterns
18. Aid decisions
19. Tell compelling stories
20. GET OUT OF TABLES
23. How did we analyze bike theft? 1. Georeferenced and cleaned the data 2. Looked for social trends 3. Looked for spatial trends 4. Looked for temporal trends
24. Georeferencing A few ways to get things on a map: 1. “Geocode” address data: 123 Main St, Lake Mills, IA, 50450 2. Georeference: pushing pins on a map 3. Use spatial data: Shapefiles, anything with lat-long
25. Cleaning the data • All records had latitude-longitude • Looked for outliers by value • Looked for outliers in space
27. 2. Looking for social trends http://factfinder.census.gov Joined ACS variables to tracts
28. 2. Looking for social trends • What’s special about spatial? Spatial Joins
32. Spatial Regressions • Which social factors correlate with bike theft? • Hard to say because extremely high spatial clustering broke the models
33. 3. Spatial Trends Understand data through aggregation at various geographic levels
34. Census Blocks Census Tracts
36. 300 foot buffers 500 foot buffers
39. 4. Temporal Trends
40. R: Time Series Decomposition
41. TIME AND SPACE! Temple U: Near Repeat Calculator
42. Impact
43. [phila street grid]
45. What are Philadelphia’s most dangerous streets?
46. PlanPhilly Identify crash hotspots: • Motor vehicle • Speeding • Aggressive Driving • Pedestrian
47. PlanPhilly Political twist for impact: • Overlay council districts, find the crash hotspots in each district • Tie into the growing Vision Zero movement and newly created urbanist PAC to influence city elections
48. Where does crash data come from? • It’s usually not open data • Reported by police in local jurisdictions, data maintained by State DOTs
53. Density Map
54. Hotspot Map
55. Hotspot maps are NOT heatmaps!
56. Hotspot Analysis • Known as the Getis-Ord GI* • ArcGIS tool in ArcMap 10.1 and later • Identifies spatially significant clusters of “high” and “low” values OR • Aggregates incident data into geographic bins for analysis
57. Spatial Join Crashes → Streets
59. PEDESTRIAN CRASH MAP
60. Why a Webmap?
61. What are we missing? • Pedestrian counts -- to create crash rate • Better way to aggregate crash data for hotspot analysis: Intersection Polygons
62. Impact PlanPhilly using this work to motivate community to demand safer, better designed streets and intersections
63. And we put the data on GitHub and OpenDataPhilly! For now...
64. Demo Time Visit this link to follow along: http://bit.ly/DGDGIS
65. Analysing the red tide with QGIS
66. Conclusion • “Spatial” opens new avenues for analysis • If you have an address or a latitude/longitude coordinate, you have spatial data • Maps are relatable and give context
67. Next Steps: GIS Tools QGIS ARCGIS
68. Next Steps: Webmapping Tools CartoDB Mapbox ArcGIS Online
69. Next Steps Learn a GIS tool + a webmap
70. Next Steps http://maptime.io/
71. Next Steps: Useful Tools Geocoding: BatchGeo Georeferencing: Geojson.io Data: Censusreporter.org Webmaps: CartoDB.com GIS: QGIS.org
72. Next Steps TAKE A GIS COURSE: Maps and the Geospatial Revolution from Penn State
73. Next Steps ASK US QUESTIONS!
74. Thanks for listening! @dmcglone @geotyler @azavea bit.ly/DGDGIS analytics.azavea.com