Brian D. Mohr and Jilan Chen Southeast Michigan Council of Governments 11 th TRB Applications...
-
Upload
ralf-simmons -
Category
Documents
-
view
217 -
download
0
Transcript of Brian D. Mohr and Jilan Chen Southeast Michigan Council of Governments 11 th TRB Applications...
Brian D. Mohr and Jilan ChenBrian D. Mohr and Jilan ChenSoutheast Michigan Council of GovernmentsSoutheast Michigan Council of Governments
1111thth TRB Applications Conference TRB Applications ConferenceDaytona Beach, FLDaytona Beach, FLMay 8, 2007May 8, 2007
SEMCOG Household Travel SEMCOG Household Travel Survey: Data Processing Survey: Data Processing
and Reasonableness and Reasonableness ChecksChecks
**DetroitDetroit
SEMCOG RegionSEMCOG Region
St. Clair
MacombOakland
Livingston
Washtenaw Wayne
Monroe
Population:Population:4.9 million4.9 million
Licensed drivers:Licensed drivers:3.4 million3.4 million
Annual VMT:Annual VMT:49 billion49 billion
Miles of road:Miles of road:23,00023,000
Communities: Communities: 234234
• Summary of data processing and Summary of data processing and reasonableness checks performed on reasonableness checks performed on 2004 household travel survey data2004 household travel survey data
• Survey background informationSurvey background information
• Calculation of survey expansion factorsCalculation of survey expansion factors
• Future initiatives and lessons learnedFuture initiatives and lessons learned
Presentation TopicsPresentation Topics
• New snapshot of regional travel New snapshot of regional travel behavior neededbehavior needed– Previous survey conducted in 1994Previous survey conducted in 1994– Shorter term enhancements planned for Shorter term enhancements planned for
four-step modelfour-step model– Possible future move to activity-based Possible future move to activity-based
modelmodel
• Opportunity to partner with MDOTOpportunity to partner with MDOT
Why Collect New Household Why Collect New Household Survey Data in 2004?Survey Data in 2004?
• Combination of two household surveysCombination of two household surveys– Michigan Travel CountsMichigan Travel Counts
– SEMCOG Travel CountsSEMCOG Travel Counts
• Survey similaritiesSurvey similarities– Consultants (MORPACE, PB, Brogan)Consultants (MORPACE, PB, Brogan)
– Activity-based survey designActivity-based survey design
– Survey methodologySurvey methodology
– Relational database structureRelational database structure
2004 Household Travel 2004 Household Travel Survey Background Survey Background
Michigan Travel Counts
SEMCOG Travel Counts
Area surveyedArea surveyed State of MichiganState of Michigan Southeast MichiganSoutheast Michigan
Total households Total households surveyedsurveyed
15,00015,000 3,8003,800
SE MI households SE MI households surveyedsurveyed
2,2002,2003,800 3,800 (for total of 6,000)(for total of 6,000)
Survey periodSurvey period Feb ’04–Mar ’05Feb ’04–Mar ’05 Oct ’04–May ’05Oct ’04–May ’05
Days surveyedDays surveyed Two daysTwo days One dayOne day
Prior residence?Prior residence? NoNo YesYes
““Stop along the Stop along the way” question?way” question?
NoNo YesYes
Transit focus?Transit focus? NoNo YesYes
• Computer-Assisted Telephone Interviewing Computer-Assisted Telephone Interviewing (CATI) logic checks(CATI) logic checks
• MORPACE post-processing checksMORPACE post-processing checks
• Parsons Brinckerhoff interim auditsParsons Brinckerhoff interim audits
• SEMCOG interim auditsSEMCOG interim audits– Review of questionable recordsReview of questionable records
– Number of persons, workers, autos per householdNumber of persons, workers, autos per household
– Distributions of trip rates and trip lengthsDistributions of trip rates and trip lengths
QA/QC Measures During QA/QC Measures During Data CollectionData Collection
SEMCOG’s Post-Processing SEMCOG’s Post-Processing Data ChecksData Checks
Database IntegrityChecks
Individual FieldChecks
Intra-RecordChecks
Inter-RecordChecks
DistributionPlots
• Checked primary keys for each tableChecked primary keys for each table
• Checked relationships among tablesChecked relationships among tables– Person Person → household→ household
– Household Household → person→ person
– Trip Trip → person→ person
Database Integrity ChecksDatabase Integrity Checks
• Determined if attribute values fell within Determined if attribute values fell within valid rangesvalid ranges
• Corrected obvious errorsCorrected obvious errors
• Found explanations for unusual errors, Found explanations for unusual errors, clarified confusing field definitionsclarified confusing field definitions
Individual Field ChecksIndividual Field Checks
• Date versus day of weekDate versus day of week• Related age fieldsRelated age fields• Related transit pass/cost fieldsRelated transit pass/cost fields• Related school variables, work variablesRelated school variables, work variables• Fields containing geocoding informationFields containing geocoding information• Origin/destination, arrival/departure fieldsOrigin/destination, arrival/departure fields• Trip-table fields related to travel modes, travel Trip-table fields related to travel modes, travel
costs, number of passengerscosts, number of passengers
Intra-Record ChecksIntra-Record Checks
• Arrival location, time compared to Arrival location, time compared to subsequent departure location, timesubsequent departure location, time
• Destination activity compared to Destination activity compared to subsequent origin activitysubsequent origin activity
• Trip characteristics for members of same Trip characteristics for members of same householdhousehold
Inter-Record ChecksInter-Record Checks
• Distributions plotted for travel times, Distributions plotted for travel times, distances, speeds, activitiesdistances, speeds, activities
• Distributions stratified by mode, Distributions stratified by mode, purpose, geographic areapurpose, geographic area
• Useful for identifying outlying dataUseful for identifying outlying data
Distribution PlotsDistribution Plots
• Overall assessmentOverall assessment– Excellent data qualityExcellent data quality
– Vast majority of checks uncovered no errorsVast majority of checks uncovered no errors
• Specific findings: database integrity, Specific findings: database integrity, individual field checksindividual field checks– Trip records discovered for “immobile” Trip records discovered for “immobile”
participantsparticipants
– Definition clarified for “stop” fieldDefinition clarified for “stop” field
Assessment of Data QualityAssessment of Data Quality
• Specific findings (intra-record, inter-record, Specific findings (intra-record, inter-record, distribution checks)distribution checks)– 22 records with incorrect day of week22 records with incorrect day of week
– 587 locations missing geocoding attributes587 locations missing geocoding attributes
– 29 records with identical arrival time and 29 records with identical arrival time and subsequent departure timesubsequent departure time
– Work trips found for households with no workersWork trips found for households with no workers
– Outliers found in some distribution plotsOutliers found in some distribution plots
Assessment of Data QualityAssessment of Data Quality
• All household locations mapped for both All household locations mapped for both MDOT, SEMCOG surveysMDOT, SEMCOG surveys
• Used to separate households in region Used to separate households in region from households outside of regionfrom households outside of region
• Used to check county attribute valuesUsed to check county attribute values
Household Geocoding Household Geocoding ChecksChecks
• Suggestions for performing specific data Suggestions for performing specific data checkschecks
• Opinion on reasonableness of basic Opinion on reasonableness of basic survey statisticssurvey statistics
• Assistance on combining two surveysAssistance on combining two surveys
• Assistance with calculating expansion Assistance with calculating expansion factorsfactors
Consultation with ParsonsConsultation with Parsons
• Concerns with second day of MDOT surveyConcerns with second day of MDOT survey– Personal trip-rates: dropped from 3.64 to 3.19Personal trip-rates: dropped from 3.64 to 3.19– Zero-trip households: increased from 8.1% to Zero-trip households: increased from 8.1% to
11.0%11.0%
• DecisionsDecisions– Combine only first day of MDOT survey with Combine only first day of MDOT survey with
SEMCOG surveySEMCOG survey– Calculate, apply expansion factors after Calculate, apply expansion factors after
combining surveyscombining surveys
Combining the SurveysCombining the Surveys
• Household size, auto ownership, number Household size, auto ownership, number of workers = 64 stratification cellsof workers = 64 stratification cells
• Spatial stratification (preferably by Spatial stratification (preferably by county)county)
• Lack of sufficient samples in some cellsLack of sufficient samples in some cells
• Balancing desire for precision, need for Balancing desire for precision, need for aggregationaggregation
Survey Expansion IssuesSurvey Expansion Issues
Calculating Expansion Calculating Expansion FactorsFactors
• Cells with insufficient samples Cells with insufficient samples aggregatedaggregated
• Initial expansion factors proposed based Initial expansion factors proposed based on experience from other urban areason experience from other urban areas
• Four-dimensional algorithm by Parsons Four-dimensional algorithm by Parsons used to calculate final expansion factorsused to calculate final expansion factors
Using Draft Expansion Using Draft Expansion FactorsFactors
CategoryCategory 2005 Households2005 Households Expanded DataExpanded Data Percent Diff.Percent Diff.
HH Size = 1HH Size = 1 519,891519,891 524,671524,671 0.9%0.9%
HH Size = 2HH Size = 2 598,931598,931 598,295598,295 -0.1%-0.1%
HH Size = 3HH Size = 3 317,948317,948 338,857338,857 6.6%6.6%
HH Size = 4+HH Size = 4+ 489,328489,328 464,667464,667 -5.0%-5.0%
Workers = 0Workers = 0 489,996489,996 490,205490,205 0.0%0.0%
Workers = 1Workers = 1 745,816745,816 755,407755,407 1.3%1.3%
Workers = 2Workers = 2 557,358557,358 563,293563,293 1.1%1.1%
Workers = 3+Workers = 3+ 132,929132,929 117,585117,585 -11.5%-11.5%
OverallOverall 1,926,0981,926,098 1,926,4901,926,490 0.02%0.02%
Using Final Expansion Using Final Expansion FactorsFactors
CategoryCategory 2005 Households2005 Households Expanded DataExpanded Data Percent Diff.Percent Diff.
HH Size = 1HH Size = 1 519,891519,891 519,891519,891 0.0%0.0%
HH Size = 2HH Size = 2 598,931598,931 598,931598,931 0.0%0.0%
HH Size = 3HH Size = 3 317,948317,948 317,948317,948 0.0%0.0%
HH Size = 4+HH Size = 4+ 489,328489,328 489,328489,328 0.0%0.0%
Workers = 0Workers = 0 489,996489,996 489,996489,996 0.0%0.0%
Workers = 1Workers = 1 745,816745,816 745,816745,816 0.0%0.0%
Workers = 2Workers = 2 557,358557,358 557,358557,358 0.0%0.0%
Workers = 3+Workers = 3+ 132,929132,929 132,929132,929 0.0%0.0%
OverallOverall 1,926,0981,926,098 1,926,0981,926,098 0.00%0.00%
• Perform additional QA/QC checksPerform additional QA/QC checks• Analyze transit-focused survey datasetAnalyze transit-focused survey dataset• Develop detailed survey analysis report Develop detailed survey analysis report
(including 1994/2004 data comparison)(including 1994/2004 data comparison)• Develop summary report (regional Develop summary report (regional
snapshot for public/media)snapshot for public/media)• Use data in modelUse data in model
Future InitiativesFuture Initiatives
• QA/QC essential from data collection QA/QC essential from data collection through post-processingthrough post-processing
• One travel day sufficient for our needsOne travel day sufficient for our needs
• GIS: useful tool for performing checksGIS: useful tool for performing checks
• Four-dimensional expansion factor Four-dimensional expansion factor calculation possiblecalculation possible
Lessons Learned Lessons Learned
Brian D. Mohr and Jilan ChenBrian D. Mohr and Jilan ChenSoutheast Michigan Council of GovernmentsSoutheast Michigan Council of Governments
1111thth TRB Applications Conference TRB Applications ConferenceDaytona Beach, FLDaytona Beach, FLMay 8, 2007May 8, 2007
SEMCOG Household Travel SEMCOG Household Travel Survey: Data Processing Survey: Data Processing
and Reasonableness and Reasonableness ChecksChecks