Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Time in HotspotBrett Lord-CastilloGIS ProgrammerSt Louis County Emergency Management                                      ...
A simple problem Given a set of designated crime “Hotspots” Find out how much time police vehicles spend in  those hotsp...
The data Automatic Vehicle Location (AVL)   Car ID   Latitude/Longitude   Timestamp Every 30 seconds per vehicle Hun...
The data-complications No specific order Transmitters malfunction (and we lose a car) Receiver malfunctions (and we los...
Manual method Convert Latitude/Longitude to XY Point table Select by Location with Hotspots polygons Export and sum up ...
Why manual doesn’t work Very time consuming, especially with manually  assigned times Have to redo selections for each d...
The Workspace
Not as complicated as it looks
Initialize and collect recordswith a blocking transformer
Inside the Loop: Three Date FormatsDay of the Week: String created by format %a (Name of day of week)Hour of Day: Number c...
Get some more informationfme_feature_type: Keep track of which points are from this week_prevcar: Car ID from the last fea...
Assign the time this return is worth
Set aside info for next feature
Spatial FilterSelect by Location, Spatial Join,and Blocking Transformer all in one
Output: Save Our Work(With fanout writer)           Week1019_1025   Pings1019_1025
Points for analysisAnd one sheet in an Excel Workbook
Last two sheets in the workbook
Outcomes
Comparison First manual attempt took over 4 hours   (To process one day)   And assumed one point = 30 seconds   Single...
Thank You! Questions? For more information:   Brett Lord-Castillo, blord-castillo@stlouisco.com   St Louis County Emer...
Upcoming SlideShare
Loading in …5
×

Determining Time in Hotspots with the Help of FME

8,967 views

Published on

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

Determining Time in Hotspots with the Help of FME

  1. 1. Time in HotspotBrett Lord-CastilloGIS ProgrammerSt Louis County Emergency Management April 10th, 2013
  2. 2. A simple problem Given a set of designated crime “Hotspots” Find out how much time police vehicles spend in those hotspots  Which hotspots?  Which days?  What hours?  Which cars?
  3. 3. The data Automatic Vehicle Location (AVL)  Car ID  Latitude/Longitude  Timestamp Every 30 seconds per vehicle Hundreds of thousands of records per day
  4. 4. The data-complications No specific order Transmitters malfunction (and we lose a car) Receiver malfunctions (and we lose hours) 5 minute stopped = stops transmitting (But GPS Drift can bring it back on) Thirty seconds is rarely thirty seconds New dataset every day (And sometimes we even lose a day)
  5. 5. Manual method Convert Latitude/Longitude to XY Point table Select by Location with Hotspots polygons Export and sum up number of returns Assume each return is 30 seconds -or- Manually assign times to each point Sum up time on all returns
  6. 6. Why manual doesn’t work Very time consuming, especially with manually assigned times Have to redo selections for each dimension  By Hotspot  By Car  By Day of Week  By Time of Day
  7. 7. The Workspace
  8. 8. Not as complicated as it looks
  9. 9. Initialize and collect recordswith a blocking transformer
  10. 10. Inside the Loop: Three Date FormatsDay of the Week: String created by format %a (Name of day of week)Hour of Day: Number created by format %H (24-hour hour of day)epoch: Number created by format %s (seconds since epoch)_totaltime: Placeholder to assign time for each GPS return
  11. 11. Get some more informationfme_feature_type: Keep track of which points are from this week_prevcar: Car ID from the last feature processed (We store this later)
  12. 12. Assign the time this return is worth
  13. 13. Set aside info for next feature
  14. 14. Spatial FilterSelect by Location, Spatial Join,and Blocking Transformer all in one
  15. 15. Output: Save Our Work(With fanout writer) Week1019_1025 Pings1019_1025
  16. 16. Points for analysisAnd one sheet in an Excel Workbook
  17. 17. Last two sheets in the workbook
  18. 18. Outcomes
  19. 19. Comparison First manual attempt took over 4 hours  (To process one day)  And assumed one point = 30 seconds  Single stat (total time by hotspot) FME ran entire week in under 30 minutes  Automated  Less assumptions  More dimensions (By Car, By Day, By Hour)
  20. 20. Thank You! Questions? For more information:  Brett Lord-Castillo, blord-castillo@stlouisco.com  St Louis County Emergency Management  @blordcastillo  https://github.com/marigolds6

×