Call Us+1 (319) 318-2400
Our last article discussed the vital role Automatic Passenger Counters (APCs) play in ridership counting for bus systems since the Covid-19 pandemic began. We also discussed the critical role that APC data plays in making strategic management decisions. While certifying the system can seem like a daunting process, the process can be streamlined with the right vendor and a qualified statistician. This article will discuss how agencies can certify their APC system using data cleansing and a calculated expansion method for reportable APC ridership data.
Why is a formal method applied to APC data before reporting? Simply put, APC data can be missing for a portion of the operated trips and tends to be slightly inaccurate, resulting from many variables including, but not limited to, operator actions, passenger hesitation, doorway transactions, etc. An agency’s APC data needs some editing with a procedural “cleansing” and “expansion” process to remove inaccurate data and re-insert statistically acceptable ridership data to account for any missing data. An NTD Qualified Statistician supports the development of the appropriate data processing methodology for each agency and applying that methodology so that cleansed and expanded data can be verified for statistical significance (note that the FTA only requires the use of a qualified statistician for identifying the expansion method for generating annual totals.).
Getting your APC system initially ready and continuing thereafter to report ridership [unlinked passenger trips (UPT) and passenger miles traveled (PMT)] involves six general steps:
1. Cleansing Raw APC Data
2. Validating the APC Data
3. Developing Expansion Methods
4. Certifying the Expansion Method for Annual Totals
5. Implementing the Cleansing and Expansion Methods
6. Maintaining the APC System
Note that the NTD has specific requirements for Validating the APC Data and Certifying the Expansion Method for Annual Totals.
Step 1: Cleansing Raw APC Data
“It is impossible for all recovered raw APC data to be usable for NTD reporting. Agencies must have their raw APC data cleansed to generate valid APC data for NTD reporting,” states Dr. Xuehao Chu, developer of the NTD Sampling Manual and a qualified statistician for NTD reporting.
NTD does not define what constitutes “valid” APC data ready for NTD reporting. The NTD simply states that “APC data should be processed to correct for anomalies,” NTD has no specific requirements on how each agency processes its raw APC data. However, how you cleanse your raw APC data should share these common features:
· Implement an established process for generating “valid” APC data from your raw APC data (typically through a cleansing procedure).
· Maintain this established process across the next three reporting years.
· Verify that the valid APC data generated from your established cleansing process meets the NTD’s validation requirements.
· Ensure flexibility in precisely what trips get thrown out or which trips are corrected for errors.
The system that maintains your APC data needs not only to identify and maintain a cleansing procedure that has been validated for NTD reporting but also needs to be nimble enough to allow for special case adjustments, for the cleaning process as part of validating the APC system (as discussed Step 2 below) in case an initial effort fails to get the APC system certified.
Step 2: Validating the APC System
Before you use the cleansed APC data for NTD reporting, you must first have your cleansed APC data validated and certified by the Federal Transit Administration (FTA) for reporting UPT, PMT, or both.
“You must meet this requirement of getting FTA certification for the first report year for which you use your APC data for NTD reporting, any mandatory recertification year (FY2022, FY2025, FY2028, etc.), and any other year during which you install “new and substantially different APC equipment,” reminds Dr. Chu.
The FTA also has detailed requirements on how to get your APC system certified:
· You must conduct a ride-check survey on a sample of operated trips during a certification year.
· The survey sample must reach a minimum sample size which is dependent on the total number of APC-equipped vehicles.
· The survey sample must account for any sampled trip being thrown out either due to not having valid APC data or not having usable ride-check data.
· The survey sample must include heavily used trips and have full coverage of all combinations of vehicle and APC models.
· The survey sample and the matching cleansed APC data need to be within 5% of each other when compared side-by-side.
It is recommended that agencies start the certification process as early as possible during a formal certification year so that agencies have time to reconcile data if the initial effort fails to pass the less-than-5% requirement for certification.
Ideally, the system that maintains your APC data supports ride-check data and provides the means necessary to compare the cleansed APC data with the survey sample to verify that the two are within the 5% variance requirement.
Step 3: Developing Expansion Methods
Once you have your APC system certified by the FTA for the first time, you will need methods to fill the ridership gap, left by trips that are without APC hardware, without raw APC data, or without cleansed APC data. Simply adding the ridership up across all cleansed APC data will greatly understate the true ridership levels. It is essential for agencies to have cleansed APC data properly expanded before being reported as valid ridership.
The expansion methods may differ for determining different measures of ridership. Some examples of ridership measures include:
· monthly total UPT
· annual total UPT and PMT
· annual daily average UPT and PMT by schedule type
The method for calculating annual totals must be certified by a qualified statistician. It is appropriate to use more than one calculation for the expansion method. It is certainly simpler to have a single expansion method for determining all ridership measures, but the feasibility depends on agency-specific circumstances and service distribution.
The expansion methods should be designed to add ridership counts to replace the missing data. However, the data does not directly “fill the gaps” of missing data points. This process is not as precise, but it is much more accurate than using cleansed APC data alone. This is due to the measured difference between the APC and ride-check data during the process of getting FTA certification of the APC system. Once certified by the FTA, the APC data after expansion according to the approved expansion methods represent the ground truth, validated APC data.
Agencies are required to describe the selected expansion method for determining annual totals in their Certification Report to the FTA.
While an APC system must be certified by the FTA at least triennially, there is no requirement on the lifespan of these expansion methods once developed. The life of an expansion method depends on the flexibility of the method calculations. The calculation development must account for expected changes in service and other unknown future conditions. Having flexibility within the expansion methods is beneficial to their longer life; the downside is the challenge of determining when conditions have changed enough to apply minor adjustments in the calculation. Examples of relevant service changes include adding regular weekend service, adding non-regular services (e.g., special-event service), changing service frequency, adding longer route patterns, allowing route deviations, changing daily service spans, etc.
Step 4. Certifying the Expansion Method for Annual Totals
You must have your expansion method for determining annual totals certified by a qualified statistician. The FTA does not prescribe specific statistician qualifications. The reporting agency must ensure that the statistician is qualified. A qualified statistician can be an in-house staff person or a hired consultant with appropriate qualifications. A statistician without transit experience may not be the best choice as your qualified statistician for measuring transit ridership.
Your statistician should be someone who has at least the following experience:
· Knowledge of FTA’s general and APC-specific reporting requirements
· APC-related technical experience
· Knowledge of transit service characteristics
· Experience with expansion modeling, or other statistical calculable modeling
Your APC certification process needs to be developed with a focus on transit to ensure that the many variables for ridership and service are considered in the development of the expansion method. This also needs to be flexible enough to address the many service changes that can occur over time.
Step 5. Implementing the Cleansing and Expansion Methods
Implementation of expansion methods can differ significantly across agencies. Many, particularly small to medium agencies, often use the methods in a spreadsheet environment like Excel. This method may save time with a smaller service profile. The drawback of this method is that it will require ongoing maintenance and review of the data to ensure the spreadsheet is maintained appropriately. On the other end of the spectrum, the largest metropolitan transit agencies may have their expansion methods coded into customized SQL scripts, and then they run the scripts when wanted for ridership reporting.
Many mid-sized and larger transit agencies apply a software application that imports the raw APC data, procedurally cleanses the data, incorporates the expansion methods, and offers the flexibility for agencies to specify the unique features of their service into the software. The challenge to this option is for the software to offer enough flexibility to accommodate the unique circumstances of your agency.
Step 6. Maintaining the APC System
The reporting agency needs to continually review the efficacy of the APC data management so that they can identify any issues or problems with the data. The best transit software provides the ability to review data collection and monitor data health. Troubleshooting expelled data is important to improve APC hardware inspections and notify staff to act before losing excessive APC data in a failing APC unit. Since APC systems generate a huge amount of data, an effective maintenance process must be automated and integrated into the data management software.
In short, if you want to fully utilize your APC System and obtain FTA certification for NTD reporting with your APC data, you need to have a process that collects, cleans, and organizes raw APC data and a process for reintroducing ridership using an expansion method. Agencies are required by the FTA to have the expansion method for generating annual totals by a qualified statistician, and it is recommended that the statistician has knowledge of NTD requirements, experience with NTD-related certification, and experience with transit data to account for service variables. And the ongoing maintenance and calculation of the APC Data for the duration of the three-year cycle should be processed by an intelligent software application that provides ongoing review and troubleshooting support.
Having your APCs certified can be a challenging process. TransTrack is here to help assist you and your agency on the way. If you would like to schedule a discussion with one of our sales consultants, please schedule it here.
-TransTrack Solutions Group-
TransTrack Solutions Group (TTSG) is home of the TransTrack Manager, a scalable enterprise data management system, serving more than 50 agencies for over 20 years. TTSG prides itself on hiring transit professionals to serve other transit professionals. TTSG can and has managed NTD ridership reporting from all sources. For more information, visit our website at https://www.transtracksystems.net/
Nathaniel Atherstone, Senior Consultant at TransTrack Solutions Group
Kelly Coughlin-Tran, Director of Marketing & New Initiatives at TransTrack Solutions Group