AVERAGE DISTANCE FOR WAGES CUSTOMERS TO TRAVEL TO NEAREST PUBLIC BUS STOP

By Dwight Danie, GIS Manager, Sifu Zhou, Senior GIS Analyst, and Osmel Lopez, GIS Technician

ABSTRACT

This study uses ArcView® and ArcView Network Analyst® to display at a US Census Tract Level, the average walking distance from a WAGES (Work and Gain Economic Self-sufficiency) customer to the closest public bus stop in Miami-Dade County, Florida. We will discuss data acquisition and processing issues, the use of Network Analyst to find the shortest distance between multiple (thousands) of points, data aggregation and presentation.

INTRODUCTION

The Community Services Planning Center (CSPC)

The Community Services Planning Center (CSPC) was established by the Florida Department of Children and Families, District 11 in Miami-Dade County Florida. Our mission is to assist the Department and the broader community with using geographic information systems (GIS) technology and research to support program planning, intergovernmental coordination and community development assistance. Throughout South Florida, CSPC delivers services to a diverse customer base, including neighborhood indicators & asset mapping, social services master plan development, benchmarking, regional and community planning, policy and program coordination, data collection and analysis.

History of Welfare Reform in Florida

The Federal Personal Responsibility and Work Opportunity Act of 1996 set in motion the present process of welfare reform in the U.S. This Act changed Aid for Families with Dependent Children (AFDC), an entitlement program, to Temporary Aid for Needy Families (TANF), a program with a five-year lifetime cash assistance limit beginning October 1998, requiring some form of work or educational activity as a condition for receiving temporary aid. [1] The State of Florida implemented welfare reform through Work and Gain Employment Self-sufficiency (WAGES) Coalitions that were organized into 24 districts.District Eleven covers the states' southernmost counties, Miami-Dade and Monroe. The District Eleven WAGES Coalition merged July 1, 2000 with the Workforce Florida Inc., a public-private partnership that:
Safe and efficient transportation services to the welfare-to-work population, along with childcare and job training, has emerged as one of the most significant challenges to welfare reform.Without adequate transportation services, welfare recipients would struggle each day to reach the job.[3]

OBJECTIVE

The CSPC was contacted in September of 2000 by workgroup consisting of members from Florida International University, Metropolitan Planning Organization for the Miami Urbanized Area and Training and Employment Council working on a project entitled "Improving Welfare-to-Work Transportation Planning". The workgroup requested a list of demographic and indicator data including the "average distance from each welfare recipient to the closest bus stop".

The CSPC receives monthly updates from the State of Florida’s computerized WAGES system. CSPC can only provide requests for confidential WAGES data at an aggregate level, (i.e. Census Tract, ZIP Code, etc.) Therefore, the CSPC must geocode participant-level, (we do this quarterly) then aggregate the data before dissemination. The objective of this project was to attach the component of the shortest walking distance to the closest public bus stop to each participant record, then aggregate the data at the Census Tract level while computing the Average walking distance for each Tract.

METHODOLOGY

DataAcquisition

The source of WAGES data is local WAGES monthly ad-hoc report. A WAGES file contains the origin addresses of approximately 11,000 participants.
Miami-Dade County Information Technology Department maintains a central repository of countywide geographic data. It provides through its Public Information Group, a comprehensive set of GIS coverages used for this project, the year 2000 Street Network and Public Bus Stops and 1990 US Census Tracts. [4]

Data Processing

The monthly WAGES file was provided to CSPC in dBase format. Two new fields were added to the file to store X and Y coordinates. To save a significant amount of time, cases were then compared to previous month’s geocoded WAGES files. The X Y coordinates for the location of previous recipients at unchanged addresses were retrieved from our data warehouse and records were updated utilizing a Microsoft Access update query with records having matching both case numbers and addresses.

Only new cases or cases that have changed address needed to be geocoded.An extract of these records were first batch-geocoded with ArcView® address style "US Streets with Zone" with Zone being the participant’s ZIP Code and with the parameters specified in Table 1.

Table 1: Batch Geocoding Parameters

Spelling Sensitivity 95
Matching Score 80
Minimum matching score to be considered candidate 30
In this project approximately 1000 records did not "batch match" nor match with previous geocoding jobs. Two GIS technicians were able to interactively geocode these records in one day using the parameters specified in Table 2.

Table 2: Interactive Geocoding Parameters

Table 2: Interactive Geocoding Parameters

Spelling Sensitivity** 80
Matching Score 80
Minimum matching score to be considered candidate 30

**During an interactive match a lower spelling sensitivity offers more candidates from which to select.

Total records 14,346
Cash/Non-WAGES 3,348
WAGES Participants 10,998
Geocoded 10,929
Un-Geocodable 69
Public Bus Stops 11,700
Street Arcs 97,000

After geocoding, the extension packaged with ArcView, "AddXY(v1.0)", was utilized to add an X-coordinate and Y-coordinate to each record. Again using a Microsoft Access update query these coordinates were then attached to the original file.

The original WAGES dBase file, with now updated X and Y coordinates, was then added to an ArcView View as an event theme.

Calculate Shortest Distance

To calculate the distance from each origin point to each destination point, we used ArcView with Network Analyst, and Shortest Network Paths® (V1.1) extensions. Shortest Network Path extension allows computations of multi-point origins to multi-point destinations.

Hardware/Software Limitations

The extension computes all possible solutions from an origin point to all destination points. Which means that for 11,000 participants (origins) finding the shortest distance to 11,000 bus stops (destinations) approximately 121,000,000 computations would be involved.

Our Dell Precision 410 NT Workstation, with dual 450 MHz processors with 128 Mb of RAM, could not handle the number of computations (i.e. stopped responding) nor accommodate the temporary files needed for this computation. Through trial and error we determined that by splitting the WAGES file and the Bus Stop file into approximately 100 record segments and 150 record segments respectively we could process each segment in about one half hour each.

Our Solution

To do this we manually selected origin points within contiguous area and converted these to a new shape file ("Origin1"). We then manually selected surrounding destinations giving enough latitude to ensure a "closest bus stop" and also converted this to a new shape file ("Destination1"). This process was done for all of the origins (WAGES participants).Eventually we had a series of files "Origin1" to "Origin116" with their companion files "Destination1" to "Destination116".

To keep track and to prevent processing a participant more than once, we created a field in the original WAGES file (SPLIT#) and using CALCULATE command, updated the field during each selection.

Calculate Shortest Distance

The next step was to run the Shortest Network Paths® extension with each segment pair using the Miami-Dade County Street Network and with the arc length field as the cost field.

Origin File = Origin1.shp
Destination File = Destination1.shp
Street Network present in View
Cost Field = Length of Arc

The extension performs three operations. It creates two shapefiles with associated tables then joins the table of the 2ndshapefile to the table of the origin table (Origin1). The first shapefile contains polylines of ALL shortest paths to ALL destinations ("Result.shp"). The second shapefile contains the shortest path to the NEAREST destination ("Result1.shp").

The Origin1 theme with its table now joined with the Result1 containing the shortest distance to the nearest bus stop was then converted to a new shapefile (WAGESdist1). This process was repeated for all Origin/Destination pairs. This part of the process required a little over 50 hours of processing time.

Once finished, all the "split" files (WAGESdist1.shp through WAGESdist116.shp were combined into one shapefile (WagesDistAll)using the Geoprocessing Wizard’s Merge function.

Attaching Boundary Data and Summarizing

Using the Geoprocessing Wizard’s function "Assign by Location", with US Census Tract Coverage in the View, census tract boundary designations (TractID) were "attached" to each WAGES participant record. The file was then converted to a new shapefile theme (WAGESShortDist) and added to the View.

A summary by Census Tract was obtained by opening the Table of the Theme WAGESShortDist, selecting the field TractID and summarizing the field while and adding an average distance calculation to the output table (TractDist.dbf).This output table was then joined to the table of the US Census Tract shapefile.
See table with average distances below.



DATA PRESENTATION

The final Census Tract table with Average Walking Distance was converted to an Microsoft Excel spreadsheet and submitted to the Metropolitan Planning Organization workgroup with a thematic map of the same data. See Map below. The data along with other data supplied by the Community Services Planning Center was incorporated into a study, "Improving Welfare-to-Work Transportation Planning", in order to apply for a Federal Transit Administration grant within the "Job Access and Reverse Commute" section. The MPO was awarded $1,000,000.00 through this grant.
See Map below.

CSPC, using ArcView Spatial Analyst, also produced further analysis by producing half-mile celled grid of the data. Image of Analysis Map.

Table by Municipality



Graph of Participants in Length Category



OBSERVATIONS

Our analysis showed that about 9% of Miami-Dade County's WAGES participants live greater than one half mile from the closest public bus stop. Studies have shown that potential transit riders would be reluctant to walk more than this distance. About 90% of participants who live further than one half mile from a public bus stop live in the peripheries of western and southern communities of the county, areas tending to be rural but undergoing development and having population growths. Close to 1% of the WAGES participants that live within urbanized communities in and around the greater Miami area live more that one half mile from a public bus stop.

GIS technology offers the ability to study other factors that need consideration in analyzing ones ability to access public transit system to travel to work. These might include the proximity of these same transit systems to daycare centers, to training centers and to jobs. And, one might also consider commute time scheduling of transit systems and bus stop accommodations (especially in South Florida), the numbers of transfers needed, and neighborhood characteristics.

SUGGESTED USES/RECOMMENDATIONS 

The Community Services Planning Center has used this same technology to calculate averages distances clients to daycare centers, and to analyze patterns of the placement of foster children within their own communities and school districts. Other ways of using this technology might be: examining regional differences in costs of care for children, or, site selection and staff allocation studies with service centers, and protective investigation units.

The utilization of GIS technology is rapidly expanding in the field Social Service planning. We envision and increasing use shortest distance analysis. The Community Services Planning Center gave the Shortest Network Paths® (V1.1) extension a thorough "workout" by processing as many points as we did. We will continue to look for suggestions for streamlining the processing multi-point origins to multi-point destinations efficiently.


[1] Federal Law Cite
[2] http://www.workforceflorida.com/wages/wfi/about/index.html
[3] http://thomas.loc.gov/cgi-bin/qiery/z?c104:H.R.3734.ENR
[4] http://www.co.miami-dade.fl.us/itd/Rates.htm.