← Go Back

Scrum Master - USC Bus Improvement Project

January 2024 - May 2024

Project Overview

The USC bus transportation system faces challenges such as low utilization by riders, potentially impacting the optimal allocation of student tuition funds for transportation. Additionally, USC Transportation lacks existing research or data collection mechanisms aimed at understanding why students choose not to utilize the bus system, implying a deficiency in understanding potential users.

Due to the nature of USC Transportation busses being funded by student tuition funds, businesses need is demonstrated through current suboptimal transportation usage. This project aims to find ways to maximize student use and ensure efficiency. Our recommendation aims to demonstrate the need for continued USC bus funding, while also allowing the University to secure future funding for AI-driven projects.

Results

TL;DR

• Poor knowledge of the busses leading to low ridership
  • Improve information access by creating social media and email presence
  • Bus stop signage
• Times and scheduling prevent ridership
  • Decrease headway to under 10 mins between busses
  • Or schedule busses to the minute
• Bus stops (location and the stops themselves) dissuade riders
  • Make bus stops more interconnected
• Future improvements
  • Tracking app
  • Improving bus stop signage, weather shielding, seating, accessibility

Role & notable accomplishments

For this project, I functioned as the Scrum Master for our team. I led communications with clients, made sure deadlines were met, ensured that requirements were gathered, and ensured that client deliverables met and exceeded quality expectations. On top of this, I contributed to every assignment and deliverable required for this project.

My notable accomplishments within this project included:

• Utilized artificial intelligence (AI) and natural language processing (NLP) to extract entities from text from our research survey.
• Developed leadership skills through spearheading this project as a Scrum Master.
• Designed a research study survey and protocol to understand lack of ridership in USC students.

Scheduling

Sprint 1:  Deliver data collection protocol to the client 

• Researched similar case studies to implement in our design
• Utilized existing data to discover focus populations for the study
• Set up data collection protocol(s)
• Published survey
• Created informative flyers and signs to incite action by users
• Ensured adequate response rate
• Delivered data collection protocol to the client

Sprint 2: Begin data analysis

• Ensured data accuracy & integrity Ensured adequate continued distribution of advertising for the data collection
• Delivered high-level analysis of collected data
• Conducted surveys
• Ensured organization of survey responses
• Delivered data collection protocol to the client

Sprint 3: Begin drafting recommendations

• Drafted reports in data analysis software (Power BI)
• Continued to ensure data accuracy & integreity
• Decided format to present final report in
• Delivered evidence of data analysis

Sprint 4: Final reccomendations

• Drafted final recommendation
• Presented final recommendation
• Ensured competion of final report on time
• Delivered final report of recommendations and data analysis

Critical Success Factors

• The compiling of a comprehensive list of recommendations on paper supported with our data
• The data should have
  • 30 datapoints supporting each conclusion (Central Limit Theorem)
  • 95% confidence interval with 5% margin of error meaning…
    • 500 survey responses that will provide insights into why there is low bus utility. We will create data-driven recommendations for next steps for the business.

Scope

In Scope:

• High-traffic low-utility populations (dorms, school buildings)

Out of Scope:

• Areas that aren’t high-traffic low-utility
• Currently-riding users

Risks

• Lack of data
• Lack of responses
• Not enough survey exposure and distribution
• Receiving adequate assistance from relevant parties
• Other major delays

Results

Overall, the study indicated multiple areas of improvement within the bus system currently. Our survey encompassed 498 responses at the end of the collection period.

Information Distribution

Through our analysis, we discovered that a majority of people do not understand how the bus system works. This was an opinion expressed evenly across all four classifications (Freshmen: n=65 or 27.54%, sophomores: n=73 or 30.93%, juniors: n=42 or 17.79%, seniors: n=52 or 22.03%). Additionally, riders who utilized the bus 1-10 times, were still almost just as likely to be as confused as those who had not ridden it at all.

This universal lack of understanding shows a clear need for an improvement in information about using the current USC transportation system.

Throughout the NLP analysis, entities about knowledge or lack of appeared frequently in multiple questions.

One area in which information could be distributed includes the University 101 class, a course most students take during their first year attending the University of South Carolina. This course provides crucial information about the University experience, and we see a large, missed potential for information about USC transportation to be included within the course content.

Another method to improve information dissemination is improving USC Transportation’s social media and email presence. Email and social media made up 47.2% of reported sources that respondents got University service information from. We believe that by improving information dissemination in these two realms, students will be better equipped with knowledge about the bus system, improving utilization.

We recommend creating a presence on at least Instagram but encourage additional social media presence on TikTok and YouTube in the future per the use habits of Generation Z and social media (Anderson, Faverio, & Gottfried, 2023). For email, we recommend at minimum a semesterly informative email about bus usage, as well as additional updates on the bus system throughout the semester. Informative or fun stories about bus drivers, frequent riders, or bus stop renovations could help students feel that their tuition money is being utilized attentively and helpfully in funding the USC bus transportation system.

Additionally, with word of mouth comprising 20.86% of the reported sources of University information, improving social media and email presence will likely lead to improved peer-to-peer dissemination of information through word of mouth.

The NLP methodology detected multiple entities mentioning information and knowledge/lack of cited as ways the bus system could be improved. Looking at other frequently detected entities in the NLP such as signage and stops, we recommend improving printed materials and signage placed at bus stops, despite this not being mentioned as a common source of information in the section specifically asking respondents about the sources of information they prefer.

Firstly, we recommend this as it will help users actively utilizing the bus system via stops to understand the capabilities the system has, without the need for internet access for those with accessibility issues. This is recommended by the National Association of City Transportation Officials, who write, “Every transit stop must include information about routes served at the stop in a clear, legible manner. Providing clear and simple information like route and system maps, schedules, expected travel times, real-time arrival times, and ridership procedures makes the system more attractive and simpler to use, and improves rider satisfaction” (National Association of City Transportation Officials, n.d.) . Adhering to these recommendations, well-designed signage can contribute to the user experience of the bus system by showing intent in design, improving perceptions of safety and usability. Furthermore, eye-catching signage may also encourage new riders to utilize the bus system by providing clear and concise information. Overall, improvements in signage can help alleviate some of the current issues with information dissemination and accessibility. Additionally, the NLP data points to the need to improve the bus stop experience (discussed below) which we feel this will help improve.

Times/Schedule

Another frequent issue appearing in the data involved the bus system’s timing and scheduling. 151 respondents (approximate 15% of responses to “What factors most discourage you from currently utilizing the USC bus system?” or approximately 30% of overall respondents) chose that lack of reliability was a deterrent to using the bus system. The sentiment was relatively even between each year classification.

In the NLP data, times, reliability, class times, and schedule appeared frequently. Many respondents indicated that one of the largest deterrents for taking the bus included a lack of alignment with their class times.

We believe that improving the scheduling of the buses may help to alleviate issues with class time alignment. Per the Texas A&M Transportation Institute, “circulator” bus routes, or routes that operate in a circulating manner around an area, are successful when the following is met:

“For neighborhood circulars, customers prefer 5- to 15-minute headways, simple routes designed to capture short trips, and good intermodal connections,” (Texas A&M Transportation Institute, n.d.) .

Decreasing the headway for current routes to fit within this schedule is a key area for improvement we see in the current bus route data. Every route except some stops in the West Campus route in the performance statistic data as of 27 has a headway greater than 15 minutes on average over all available times. Implementing routes that arrive/depart every 5 to 15 minutes can help improve the timing issue frequently reported by respondents. On USC’s campus, this could look like an increased number of busses per route to decrease the arrival/departure time with visibility on mobile apps (including Google Maps/Apple Maps), or even live-updated electronically at each stop through smart bus stops. The headway time would likely need to be 10 minutes or less based on the size of campus and “walking/faster to walk” being cited as a common reason to utilize other modes of transportation in the NLP data. One respondent said,

“It would be beneficial to actually have the busses be at the stop when they should be. I don’t have time to wait for a bus that evidently doesn’t show up.”

User frustration with headway and wait times is certainly an area of improvement we recommend fixing to garner more bus usage.

Alternatively, USC Transport and COMET could modify the University bus routes to operate on a fixed-route service. In this version of operation, buses operate on predetermined paths and adhere to scheduled times (Transloc, 2022) . This could also help alleviate timing issues by offering predictability, allowing for students to plan their commute to their next location after classes.

Additionally, USC Transport could encourage the positioning of buses at key locations during common class exchange times. While this may not directly affect headway, it would likely improve the utilization of the buses by more students overall.

Bus Stops

A majority of respondents reported no opinion (n=144 or 37.40%) on the overall state of the bus stop stops in terms of accessibility, amenities, cleanliness, and safety. This is followed by respondents reporting satisfaction with the current bus stops (n=131 or 34.02%), respondents reporting unsatisfaction with stops (n=77 or 20%), those reporting high satisfaction with stops (n=18 or 4.67%), and those reporting high unsatisfaction with stops (n=15 or 3.89%).

The NLP analysis indicates frequent mentions of the bus stops when prompted about desired improvements and what factors contribute to the choice to use non-bus transportation to campus.

Taking a closer look at the answers mentioning bus stops, trends included requests of adding more stops (especially near student housing off campus or on campus outskirts), not understanding which stops serve which routes, adding environmental protection and seating to stops, adding signage to stops, and buses not being present at the stops. Examples include the following responses:

“More routes with stops directly in front of student housing. Student discounted fares. More sidewalks along roads to access bus stops.”

“I would only use it to go far distances like across campus. Make sure all stops have rain covers and have busses come more often.”

“Some bus stops outside of campus and farther from the city are also just on the side of the road with no side walk so a lot of bus stops aren’t accessible.

We recommend changes to the bus stops to improve the overall experience, some of which have already been discussed prior. First, improvements in signage at bus stops can help to improve the overall user experience and accessibility when utilizing the bus system. We also recommend improvements in the shelters of the bus stops to improve the waiting experience. Many bus stops on campus currently do not have seats, rain shields, or sun shields. When paired with the current long headway times, the experience of waiting at stops for the bus is negatively impacted, especially during inclement weather or for individuals with accessibility needs.

By improving the structural deficits in the bus stop experience, improvements in the visibility of bus system usage and word of mouth are sure to follow. One respondent wrote, “For me personally, if the bus stop was closer to my apartment then I would use it. I would also prefer if I saw many other students use it,” further proving the social need for visible usage paired with word-of-mouth spread of information mentioned before.

Second, we also recommend revitalizing the stops on bus routes to connect more locations at the edges of campus to classrooms in the center, including parking garages, close to off-campus housing such as Empire and Greene Crossing, and the Gambrell area of campus. For example, the Red Line could easily be modified to have a stop near Green Quad/Swearengin and Bull Street Parking Garage to connect as far as Five Points to the furthest South classrooms on campus to the furthest North classrooms, as well as to central campus.

Responses mentioning “stops” also frequently mentioned wait times at stops, though we feel the wait times could be addressed by addressing bus frequency and schedule as mentioned in the “Times/Schedule” action above.

Additional Improvements

There were frequent mentions of the Transloc app in the NLP data over all long-answer questions data (n=26), indicating issues with the application. We recommend future improvements to the application be made after other updates to the bus system are implemented. We believe that allocating time and resources towards enhancing other aspects of the bus system would yield greater benefits compared to further investment in the application currently.

NLP data indicates that night is mentioned most frequently when asked about how times and days may affect bus use (n=10). Responses were mixed between indicating hesitance to use the bus at night, while also a desire for night-time routes from classes. Improving bus stops can likely lead to an increased sense of safety. A 2018 study in Sweden found that the three most important factors promoting feelings of safety at stops are shelter, natural surveillance or visibility, and real-time information provision (Abenoza, Ceccato, Cats, & Susilo, 2018). Addressing some of these elements at USC to improve safety perceptions could include adding additional lighting, adding environmental protection, and improving mobile app tracking of the buses as per recommendation from a 2018 study that said improving shelter, natural surveillance, and real-time information provision can improve overall feelings of safety (Abenoza, Ceccato, Cats, & Susilo, 2018). With 4890 riders at night time (8pm-4am) over 86 days, there is proven existing demand that could be expanded upon. Additionally, overall improvements in the utilization of the bus system could improve safety concerns at night by increasing the population of ridership.

Limitations

The team primarily faced limitations with the technology responsible for the analyzing of the data. For instance, with Power BI one of the main obstacles for the team was the inability for multiple people to work on one report. Another problem with Power BI was the lack of a built-in data cleaning solution. Meaning, all the data had to be manually cleansed before it was processed into BI for analysis. The team also saw a few issues with our natural language processing. A big issue we had was spaCy NLP missing some of the entities. For example, it missed 3 instances of accessibility/inaccessibility in the “Additional comments” question. Additionally, context of the entities still has to be determined using human annotation when using spaCy NLP. Therefore, it is possible our team missed some context of the entities when analyzing their contexts. Future USC Transportation studies could use the information generated in this study as training data for another survey, creating an annotated dictionary to improve NLP entity recognition in the context of improving public transport and surveying students.