1. Motivation

OP CONSTRAINTS

Lowest commercial speed registered in 2024

Alemão (2025); Soldado (2024)

NEGATIVE PERCEPTION

Bus is perceived as late and unreliable

Ramos et al. (2019); Rocha et al. (2023); Ribeiro et al. (2020)

LOW MODAL SHARE

Public transit share is only 16.2%

INE (2021)

It is crucial to study solutions to improve its efficiency and reliability

1. Motivation + 2. Literature Review

How are these problems being addressed?

Literature focuses on passenger satisfaction and performance-driven indicators

1. Motivation + 2. Literature Review

How are these problems being addressed?

Literature focuses on passenger satisfaction and performance-driven indicators

Drivers are overlooked by most studies

1. Motivation + 2. Literature Review

How are these problems being addressed?

Literature focuses on passenger satisfaction and performance-driven indicators

Drivers are overlooked by most studies

1. Motivation + 2. Literature Review

How are these problems being addressed?

Literature focuses on passenger satisfaction and performance-driven indicators

Drivers are overlooked by most studies

Considered as an output of the system

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology


Mixed-methods approach to capture both the 📊 explicit and 🧠 tacit knowledge

3. Methodology

State-of-the-Practice

Structured interviews with operational control professionals from several cities

3. Methodology

Case Study

Lisbon public bus operator, part of Navegante multimodal system

102

Routes

1,658

Drivers

100 M

Pax/Year

3. Methodology

Case Study

Carris faces growing challenges caused by 🚗 conflicts with private vehicles and
🚧 construction works (Carris 2024)

Demonstrated by Carris and computed indicators
Yellow indicators were computed by the author, considering operational events for May 2025

13.71

Km/h

7.5 %

Bus bunching

4.9 %

Short turns

95 %

Regularity

3. Methodology

Focus Group Method and Analysis

1 session with 8 drivers, sampled by operator with diversity criteria

3. Methodology

Focus Group Method and Analysis

Held at Pontinha Garage, lasting 90 minutes

3. Methodology

Focus Group Method and Analysis

Moderated by research team, following a semi-structured discussion
(Hennink et al. 2020; Tracy 2013)

3. Methodology

Focus Group Method and Analysis

Content analysed and reported based on thematic analysis
(Braun and Clarke 2006)

3. Methodology

Exploratory Data Analysis

Dataset from XTraN, the operational assistance system used by Carris

👮‍♀️ Controllers




🧑‍✈️ Drivers

3. Methodology

Exploratory Data Analysis: Messages

Text communication between drivers (on-board tablet) and controllers (Operational Control Center)

📅 From Jan to May 2025 (132 days) 🖇️️+ 270k records

{
  "DriverMec": "189944",
  "DriverName": null,
  "DriverStationCode": "3739",
  "SenderID": 2937,
  "OperatorName": 23554,
  "OriginType": 1,
  "GeometryPoint": "{\"type\":\"Point\",\"coordinates\":[-9.1473,38.74355]}",
  "GpsQuality": 3,
  "Id": 9736723,
  "Date": "/Date(1749431120000)/",
  "ProcessedDate": "/Date(1749431130883)/",
  "VehicleNumber": 796,
  "VehiclePlate": "",
  "BusNumber": 1761,
  "RouteNumber": "207",
  "PlateNumber": "1Z207",
  "PlateNumberText": "207-1",
  "MessageText": "ACIDENTE ENTRE TERCEIROS",
  "Status": 4,
  "FirmGroup": null,
  "FontColor": "#2b2b2b",
  "BackgroundColor": "#d9cbfb"
}

3. Methodology

Exploratory Data Analysis: Messages

🧑‍✈️ 2.78 messages per driver/day (SD = 2.35)

3. Methodology

Exploratory Data Analysis: Messages

🧑‍✈️ 2.78 messages per driver/day (SD = 2.35)

67.51% are procedural

3. Methodology

Exploratory Data Analysis: Messages

🧑‍✈️ 2.78 messages per driver/day (SD = 2.35)

67.51% are procedural

👮‍♀️ 61.4% by controllers

3. Methodology

Exploratory Data Analysis: Messages

🧑‍✈️ 2.78 messages per driver/day (SD = 2.35)

67.51% are procedural

👮‍♀️ 61.4% by controllers

⚠️ 71.8% are free text messages

3. Methodology

Exploratory Data Analysis: Messages

👮‍♀️ 61.4% by controllers

⚠️ 71.8% are free text messages

Text mining processing using Latent Dirichlet Allocation topic modeling and bigrams analysis
(adapted from Valença et al. 2023)

Spatial dynamics of messages categories to reveal geographical and temporal patterns

3. Methodology

Exploratory Data Analysis: Events

Logs of automatic vehicle positions

📅 May 2025 (31 days) 🖇️️+ 32M records

{
  "Id": 2545991349,
  "DriverName": null,
  "DriverMec": "189294",
  "DriverStationCode": "3779",
  "BusNumber": "4627",
  "VehicleNumber": 168,
  "PlateNumber": "4Z736",
  "PlateNumberText": "736-4",
  "RouteNumber": "736",
  "Direction": "DESC",
  "VoyageNumber": 4,
  "TotalKm": 702783,
  "EventTime": "/Date(1749457228000)/",
  "EventReceptionTime": "/Date(1749457291137)/",
  "GpsQuality": 3,
  "Latitude": 38.75392,
  "Longitude": -9.15257,
  "Speed": 14,
  "BusStopNumber": 5704,
  "Description": "Detectada a paragem Campo Grande - Av. Brasil"
}

⚠️ Normalization issues

Algorithm to identify events associated to each trip planned

3. Methodology

Exploratory Data Analysis

4. Empirical Findings and
Thematic Analysis


4. Empirical Findings and Thematic Analysis

State-of-the-Practice

Interviews with controllers from 4 cities

Lisbon 🇵🇹

Porto 🇵🇹

Braga 🇵🇹

Portland 🇺🇸

4. Empirical Findings and Thematic Analysis

State-of-the-Practice

3/4 use on-board operational assistance systems

Portland presents most comprehensive setup

Portland reports highest engagement

4. Empirical Findings and Thematic Analysis

State-of-the-Practice

3/4 use on-board operational assistance systems

Portland presents most comprehensive setup

Portland reports highest engagement

All value previous driving experience for controller recruiting

Only Lisbon and Portland report structured participation

Portland provides feedback through interactive dashboard

4. Empirical Findings and Thematic Analysis

State-of-the-Practice

3/4 use on-board operational assistance systems

Portland presents most comprehensive setup

Portland reports highest engagement

All value previous driving experience for controller recruiting

Only Lisbon and Portland report structured participation

Portland provides feedback through interactive dashboard

None reported to include drivers in planning team

Most have channels for driver input

Portland stands out with direct channel

4. Empirical Findings and Thematic Analysis

State-of-the-Practice

3/4 use on-board operational assistance systems

Portland presents most comprehensive setup

Portland reports highest engagement

All value previous driving experience for controller recruiting

Only Lisbon and Portland report structured participation

Portland provides feedback through interactive dashboard

None reported to include drivers in planning team

Most have channels for driver input

Portland stands out with direct channel

Overall, drivers’ experience is recognized as valuable, but structured inclusion of their knowledge in monitoring and planning remains limited in Portugal 🇵🇹

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

18 codes, grouped in 6 categories

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

Connect passengers from A to B on time, with comfort, safety and no complaints

Drivers miss guidance on how to balance competing objectives

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

 

Illegal parking

Bus lanes and stops occupied by ride-hailing and private vehicles

Touristic pressure

 

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

 

XTraN lacks updates, feedback and conection reliability

Highway patrol discourages road code infractions

 

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

 

Unrealistic schedules (traffic, urban changes, touristic demand)

Ticketing improvements with contactless

 

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

 

Network outdated with urban development + missing toilets

Intersections and traffic lights as bottlenecks

Unattractive career

 

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

Drivers express pride in Carris mission and their role

But also a sense of not being considered in decisions processes

Factor improves operational performance | Factor damages operational performance

4. Empirical Findings and Thematic Analysis

Focus Group Thematic Analysis

Drivers tacit knowledge goes beyond their functional role and hold valuable
system-wide perspectives

Potential to complement explicit knowledge of planners

4. Empirical Findings and Thematic Analysis

Focus Group Ranking Analysis

Ranking of 11 operational factors prioritize Tactical Planning (Borda Score = 61.4) and Strategic (Borda Score = 54) to Operational (Borda Score = 49.7)

4. Empirical Findings and Thematic Analysis

Focus Group Ranking Analysis

Ranking of 11 operational factors prioritize Tactical Planning (Borda Score = 61.4) and Strategic (Borda Score = 54) to Operational (Borda Score = 49.7)

Further supports that drivers tacit knowledge goes beyond their functional role and hold valuable system-wide perspectives

4. Empirical Findings and Thematic Analysis

Comparing Findings With Previous Studies

Results are aligned with the literature

Factor Subfactor Source
Weather Impact on operations 📖 👥
Other vehicles Traffic 📖 👥
Passengers Irregular loads; Traveler behaviour 📖 👥
Service planning Network design; Schedule quality 📖 👥
Vehicle Modernization 📖 👥
Stop design Location; Traffic flow segregation; Number of berths 📖 👥
Stop operation Boarding conditions; Schedule design; Traffic lights; Payment system 📖 👥
Bus lane Impacts; Design 📖 👥
Drivers Experience; Stress; Fatigue and drowsiness 📖 👥

📖 Literature Review | 👥 Focus Group

4. Empirical Findings and Thematic Analysis

Comparing Findings With Previous Studies

Results are aligned with the literature and extend it

Factor Subfactor Source
Weather Impact on operations 📖 👥
Other vehicles Traffic 📖 👥
Passengers Irregular loads; Traveler behaviour 📖 👥
Service planning Network design; Schedule quality 📖 👥
Vehicle Modernization 📖 👥
Stop design Location; Traffic flow segregation; Number of berths 📖 👥
Stop operation Boarding conditions; Schedule design; Traffic lights; Payment system 📖 👥
Bus lane Impacts; Design 📖 👥
Drivers Experience; Stress; Fatigue and drowsiness 📖 👥
Service operation Operating assistance system; Enforcement ➖👥
Governance Working conditions; Company commitment on stakeholder engagement ➖👥

📖 Literature Review | 👥 Focus Group

5. Operational Data
Analysis Outcomes


5. Operational Data Analysis Outcomes

Free Text Messages Categorization

LDA produced 12 topics that were manually reviewed and interpreted

5. Operational Data Analysis Outcomes

Free Text Messages Categorization

Complemented by bigrams that reinforced and extended previous findings

5. Operational Data Analysis Outcomes

Free Text Messages Categorization

Text mining identified 7 new categories and mapped to 7 pre-existing ones

⚠️ 71.8% are free text messages

64,522 re-categorized

10% free text messages

5. Operational Data Analysis Outcomes

Free Text Messages Categorization

Text mining identified 7 new categories and mapped to 7 pre-existing ones

⚠️ 71.8% are free text messages

64,522 re-categorized

10% free text messages

Suggests that XTraN system is misaligned with operational needs

5. Operational Data Analysis Outcomes

Messages Spatial Dynamics

5. Operational Data Analysis Outcomes

Messages Spatial Dynamics

Wheelchair passengers 🦽

5. Operational Data Analysis Outcomes

Messages Spatial Dynamics

Wheelchair passengers 🦽

5. Operational Data Analysis Outcomes

Messages Spatial+Temporal Dynamics

 

Full bus 👥

Ticket fraud 🚨

 

5. Operational Data Analysis Outcomes

Messages Spatial+Temporal Dynamics

 

Full bus 👥

Ticket fraud 🚨

 

There is a high potential to convert tacit observations into explicit knowledge to support decision-making

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

Most categories tend to be sent during trips that take longer than scheduled

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

Most categories tend to be sent during trips that take longer than scheduled

Delayed arrivals are related to traffic, crashes, crowding, reserved and short turns

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

Most categories tend to be sent during trips that take longer than scheduled

Delayed arrivals are related to traffic, crashes, crowding, reserved and short turns

Overtake seems to be effective on restoring schedule regularity, unlike control driving

5. Operational Data Analysis Outcomes

Linking Messages to Trip Performance

Considering average values, in minutes

Message N Departure delay Travel time diff Arrival delay
Full 6621 0.3 3.4 3.7
Proceed 2606 -0.1 2.1 2.0
Short turning 2043 -0.3 5.8 5.5
Traffic 1782 0.4 4.6 4.9
Overtake 173 0.4 -0.3 0.1
Crash 167 0.5 3.5 4.0
Schedule warning 61 -0.4 3.4 3.0
Control driving 52 -0.7 -4.2 -5.0
Reserved 4 0.5 4.0 4.5

Messages patterns could be incorporated in automated real-time alarms or responses to enhance operational response time and mitigate impacts of disruptions

6. Conclusion


6. Conclusion

Main Findings

Results demonstrate that drivers have relevant insights about all planning levels, which can meaningfully contribute to decision making

6. Conclusion

Main Findings

Results demonstrate that drivers have relevant insights about all planning levels, which can meaningfully contribute to decision making

There is an untapped potential to turn experience into organizational knowledge, that can ultimately improve service reliability and overall system performance

REDUCE OP CONSTRAINTS

NEGATIVE IMPROVE PERCEPTION

LOW INCREASE MODAL SHARE

6. Conclusion

Main Findings

Results demonstrate that drivers have relevant insights about all planning levels, which can meaningfully contribute to decision making

There is an untapped potential to turn experience into organizational knowledge, that can ultimately improve service reliability and overall system performance

Technological innovation alone is not sufficient without serious institutional commitment to involve stakeholders and promote continuous improvement

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for particited decision framework

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for participated decision framework

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for particited decision framework

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for particited decision framework

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for particited decision framework

6. Conclusion

Policy recommendations for Carris


  1. Improve messages categories to meet operational needs
  1. Improve messages functionality to increase usability
  1. Incorporate on-board automatic detection of recurring disruptions
  1. Evolve XTraN from reactive to predictive decision-support tool
  1. Create structured channels for driver feedback
  1. Institutional commitment for particited decision framework

6. Conclusion

Methodological Contributions

Replicable framework to assess how tacit and explicit knowledge coexists within organizations

Code available at gmatosferreira/Through-bus-drivers-eyes

Acknowledgements


 

 

References

References

Alemão, Samuel. 2025. “A Passo de Caracol. Autocarros Da Carris Com a Mais Baixa Velocidade de Sempre.” In PÚBLICO. https://www.publico.pt/2025/04/15/local/noticia/passo-caracol-autocarros-carris-baixa-velocidade-2129894.
Braun, Virginia, and Victoria Clarke. 2006. “Using Thematic Analysis in Psychology.” Qualitative Research in Psychology 3 (2): 77–101. https://doi.org/10.1191/1478088706qp063oa.
Carris. 2024. Relatório e Contas 2024. https://www.carris.pt/media/ikvdezmc/relat%C3%B3rio-e-contas-carris-2024.pdf.
Hennink, Monique, Inge Hutter, and Ajay Bailey. 2020. Qualitative Research Methods. SAGE.
INE. 2021. “Indicador: População residente que vive no alojamento a maior parte do ano (N.o) por local de residência à data dos censos [2021].” https://tabulador.ine.pt/indicador/?id=0011704.
Ramos, Sara, Paula Vicente, Ana M. Passos, Patrícia Costa, and Elizabeth Reis. 2019. “Perceptions of the Public Transport Service as a Barrier to the Adoption of Public Transport: A Qualitative Study.” Social Sciences 8 (5): 150. https://doi.org/10.3390/socsci8050150.
Ribeiro, Paulo, Fernando Fonseca, and Paulo Santos. 2020. “Sustainability Assessment of a Bus System in a Mid-Sized Municipality.” Journal of Environmental Planning and Management 63 (2): 236–56. https://doi.org/10.1080/09640568.2019.1577224.
Rocha, Hudyeron, Manuel Filgueiras, José Pedro Tavares, and Sara Ferreira. 2023. “Public Transport Usage and Perceived Service Quality in a Large Metropolitan Area: The Case of Porto.” Sustainability 15 (7): 6287. https://doi.org/10.3390/su15076287.
Soldado, Camilo. 2024. “Em Mais de 15 Anos, Autocarros de Lisboa e Porto Nunca Estiveram Tão Lentos.” In PÚBLICO. https://www.publico.pt/2024/06/20/local/noticia/15-anos-autocarros-lisboa-porto-tao-lentos-2094651.
Tracy, Sarah J. 2013. Qualitative Research Methods: Collecting Evidence, Crafting Analysis, Communicating Impact, 2nd Edition Wiley. 1st ed. Wiley-Blackwell. https://www.wiley.com/en-be/Qualitative+Research+Methods%3A+Collecting+Evidence%2C+Crafting+Analysis%2C+Communicating+Impact%2C+2nd+Edition-p-9781119390800.
Valença, Gabriel, Filipe Moura, and Ana Morais de Sá. 2023. “How Can We Develop Road Space Allocation Solutions for Smart Cities Using Emerging Information Technologies? A Review Using Text Mining.” International Journal of Information Management Data Insights 3 (1): 100150. https://doi.org/10.1016/j.jjimei.2022.100150.