InnoConnect: Big data analysis
- 1. Big Geo-Data Analysis 'Transforming DATA into INFORMATION' Jiří Bouchal Big & Open Data and Open Software Workshop January 2018 Novotného lávka, Praha, Czechia
- 2. Transforming Data into Information „The purpose of visualization is insight, not pictures“ - Ben Shneiderman, 1994
- 3. 3 Acting on data-driven insights There is a big difference between raw data and information
- 4. Big Data Visualization with WebGLayer • JavaScript library (open source) • Big data visualisation through web browser • Visualisation of point and linear data
- 5. Features • Interactive visualisation of up to 1.5 mil data records • Response time below 100ms • Compatible with 3rd party map API (Leaflet, Google maps) ✓ Interactivity ✓ Scalability ✓ Responsiveness ✓ Web-based, no specific GIS software
- 6. Outperforming Competitors 6 Test Results ● Datasets of up to 1.5M points tested on usual laptop ● ArcGIS Online fails to load more then 100K points ● Google Maps API loads 1M points after 6.5s. Leaflet after 2s ● WebGL loads 1.5M points in 0.1s
- 7. 7 UK 1.5 million traffic accidents
- 8. Fast Interactions Linked views: dynamic data filtering and zooming
- 9. Live Demonstration Flanders 9 Flanders ● identification of traffic accident hotspots nearby schools on weekdays between 7- 9h ● Discover deadly accidents in the Antwerp area by applying polygon filter(amount, locations, times) http://innoconnect.net/apps/flanders- traffic-accidents/
- 10. Live Demonstration: Traffic 10 Turnhout, Belgium ● Traffic flow recorded by one camera ● Driving patterns among cameras http://innoconnect.net/development/mac q-uc3/
- 11. Live Demonstration Chicago 11 ● See the distribution of crimes over time and space (e.g. prostitution, homicide, interference with p.off.) ● Chicago (600 homicides in 10 month/population 2.7M) vs. Czechia 150 homicides a year/10M) ● Prostitution 600 (all arrested) vs. homicides 600 (only 100 arrested) ● Discover districts and streets with high risk of certain crime type (car theft, drugs, burglary etc.) ● Identify patterns: e.g. weapons, narcotics, gambling and prostitution in same areas ● 50k+ thefts. Most in downtown (6k+) ● Battery, robbery and assault (40+10+15k) – west and south ● Identify downtown nightlife areas with highest criminality
- 12. Wide Range of Potential Use 12 ● Criminality data - discovering risky areas ○ Parking offenses ○ Thefts, burglaries, robberies ○ Gambling ○ Motor vehicles thefts ○ Public peace violations etc. ● Traffic accidents analysis
- 13. Wide Range of Potential Use 13 ● Traffic intensity data from road sensors and cameras ● Parking occupancy analysis and availability prediction ● Water management companies ○ Water usage ○ Water losses caused by pipe leaks and failures
- 14. Next Developments 14 Traffic flow visualisation - Flanders Traffic accidents - Belgium Traffic intensity - Plzeň ? Criminal offenses - Plzeň
- 16. New Features 16 New Colour Schemes for Heatmaps