1 / 29

O/D Applications from Smart Card Data

O/D Applications from Smart Card Data. Jesse Simon, Ph.D. simonj@metro.net. The O/D we are targeting. Origin = the initial boarding stop of a linked transit trip. Destination = the final alighting stop of a linked transit trip.

Download Presentation

O/D Applications from Smart Card Data

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. O/D Applications from Smart Card Data Jesse Simon, Ph.D. simonj@metro.net

  2. The O/D we are targeting • Origin = the initial boarding stop of a linked transit trip. Destination = the final alighting stop of a linked transit trip. • Only boardings and alightings on transit vehicles are being measured • The transit O/D not the actual O/D. • O/D involves linked trips: while only 1 vehicle can be involved (a 1-link trip), 2 or more vehicles are often involved in the trip. • On/Off counts do not tell us this info since we need to know where an individual patron starts and ends his trip, which may involve transfers.

  3. WE KNOW WHAT A SINGLE CARD USER DOES ON ANY GIVEN DAY. • Every Smart Card Transaction is date, time and geo-stamped • Includes the Card ID number, which is stored on the farebox computer and downloaded to a database • Smart Card transactions now constitute 45% of all Fare transactions, and is growing

  4. Differs from BART & WMATA • BART & WMATA have entry/exit system • They know actual O/D • We have entry-only farecards (smart cards) • We must infer Destination • We know the origin: first boarding stop of initial link • We infer the final link destination by matching trips • On any given day, the card’s later trip’s initial boarding stop is the inferred destination of the earlier trip. • Vice-versa: later trip destination is earlier trip’s first stop

  5. Hot Topic • NYC, Chicago, Ottawa, London, Sao Paulo all have pilot programs. • Benefits: • Much larger volumes than On Board Surveys • Low marginal cost because already collected • Available in weeks rather than a year • Allows ongoing Time Series comparisons • No problems of OBS self response • Accuracy and self-selection problems ameliorated

  6. O/D Inference Efforts Compared Other Cities’ Efforts tied to Modeling • Assume Day’s First & Last Trip match • Use NYC finding of 90% accuracy as reasonable approximation for modeling Our #1 audience, schedulers and schedule planners, demand greater precision • Every matched pair tested (loss of quantity, gain in validity)

  7. Preview: What O/D shows • Origin/Destination (O/D): Where patrons’ trips begin and where they end • The last map showed 3 O/D patterns from TAP card data: (text matches “desire line” colors) • 3rd/Vermont: Local catchment area • El Monte Station: Wide catchment area but mostly from SG Valley and Downtown LA • Metro Center: Extremely wide catchment area with heavy O/D along Rail and Harbor Freeway corridors

  8. Study part of validation project Error Check, Troubleshoot, Correct: • Apply diagnostic reporting & liaison with Maintenance that made APC work better at LACMTA than elsewhere • Develop prototype applications that identify problems • Project now developing reports & liaison • Troubleshooting methods/reporting subject of another paper.

  9. Two Tasks: • Form Linked Trips • Match Linked Trips to Infer O/D A successful Match: First/Last stops match • Outgoing 3-Link AM trip (black) • Incoming 2-Link PM trip (brown)

  10. 1st Step: Defining a Linked Trip • Before Matching Linked Trips, form them • On any given date on any given card there may be one or many fare transactions • Each transaction represents a boarding which, in this context, we call a “link” • The question is: which links become part of a linked trip • The Solution:

  11. Forming Linked Trips • What Other Agencies Do • Use “fixed temporal thresholds” between transactions • Less than 30 or 90 minute elapsed time between boardings, or less than 1 hour wait time at stops • Criticized for not accounting for variation in trip length and service levels • Use “Spatial-Temporal” path (Ottawa demo project) • Uses alighting time of first link destination stop and walk speed of 2.7 mph + 5 minutes to next stop. If boarding on the 2nd link can be made in that time period then link is assumed. (Chu & Chapleau)

  12. Forming Linked Trips • What was done here • Link is part of linked trip if time elapsed between successive boardings greater than 3 mph • Retains Spatial-Temporal context where “miles” is spatial and “per hour” is temporal • Processing pragmatics: • Ottawa viewed thousands of trips using multiple data sources, including referring to other passenger arrival times; • We will be regularly looking at millions of trips so we want to keep it to one data source (TAP Cards), one user at a time • 3 mph is more a function of service provided than patron ability: • Very few instances where Metro service, including headways, was lower than 3 mph between any two connecting lines at any two stops (not even downtown LA) • Other agencies may need lower mph to capture links

  13. 2nd Step: O/D Matching

  14. Matching Stop Areas Matching Lines connected to Stop Areas based on geostamp of boarding Location • Actually where the origin and destination is • Eliminates database Line attribution error • Foreign Line recording • Trunk vs. Branch Line designation

  15. Line Destiny Report Linked Trip generation is important in itself • Even without inferring final destination • Product: Report that rank orders transfers to Lines from any given Line • Much more complete dataset than O/D because unmatched trips are included • 75% of trips put into Linked trips (will be much higher in future); compares to 38% O/D matches. • Fewer inferences involved

  16. Selection Bias • The 38.3% matching among Fare Card users points to selection bias • Appeared dramatically in study of a commuter college area, where home-school-work-home tours could not be matched. • But generally, the core group of multi-use fare card users are those who use Metro 5 days a week. • 5 day a week users are 82% of all users, and 82% of their trip productions are either home-work or home-school • Conclusion: biased sample of total transit travel but representative of core travel.

  17. General Findings • The basic travel pattern is the 1-link trip: 57.3% are 1-link • Linkage from any given Line is widely distributed among transfer points • Median highest % of trips destined to another Line is 4%. • Only 6 Lines have 10% of patrons destined to another specific Line • No Metro Rapid Line has 10% of its patrons transfer to the Local Line on the same corridor • Contradicts Original assumptions

  18. What a Map Application Looks Like See accompanying Map Destinations of Trips Originating on Line 901 • The Orange Line distributes people to: • Other Orange Line Stations (most frequent) • The Red & Purple Line corridors • Downtown LA • Hollywood • Small concentration in Westwood • Many N/S Corridors throughout the SFV, especially Van Nuys Blvd. • It does not distribute to the Blue or Gold Lines • This map does not show trips where Orange Line is an intermediate link on a 3 link trip.

  19. Application: Travel to a specific area Background: Original research was on shortening Line 761 route. • Question arose as to O/D for people in Westwood area generally. • In this case “Westwood” was defined as areas in or near Westwood that people on Van Nuys corridor traveled to • All origins, including those outside SFV, included in the accompanying map

  20. Use of Census Tracts in Map • Census Tracts are color coded to show intensity of travel (number of trips) to each of them. • By coloring tractfor destinations and tract interiors for origins, origins and destinations can be compared. • Census Tracts have demographics attached to them. Travel behavior can then be tied to demographics. • Census Blocks, Block Groups, TAZs or any other geographic area can be used outlines

  21. Findings • There are two destination tracts in Westwood that dwarf all the others: UCLA (238 trips) and the tract along and south of Wilshire by Westwood Boulevard (128 trips) • An optimal stop on the subway to the sea would be on Wilshire between these two census tracts. • Most of the origin tracts lie on three main corridors: Van Nuys (with a short jog on Ventura), Wilshire/Whittier, and Sunset • The heavy origins are as far north as Nordhoff on Van Nuys • The heavy origins stretch very far to the east on both Wilshire/Whittier and Sunset • They trace out a strong path for potential corridors of the subway to the sea. • All three corridors represent some long trip-making. • The UCLA tract has the most origins, which indicates travel within Westwood. • These are short trips.

  22. Next Steps: at LACMTA • Work with UFS on data errata and, if needed, structure requirements • Rewrite programs to retain information on intermediate trips • This will allow additional kinds of maps, such as O/D of trips where Line 901 is an intermediate link • Also important to Modeling as potential data client: critical paths and links per trip • Create data structures and routines for regular processing of O/D datasets for queries and mapping

  23. Next Steps: Research Community • Coordinating with On/Off data • Coordinating data from automatic sources (Fare Card Data, On/Off counts) with intentionally developed survey data • Recognizing, exploiting and combining their strengths and weaknesses • Calibrating or supplementing massive datasets with survey data

More Related