General Assembly 2017, Stockholm, Sweden Development of predictive models for public transport operations to improve user information, indicators and real-time forecast for operators and transit agencies motivacion Ricardo Giesen, Juan Carlos Herrera, Hans-Albert Löbel, Felipe Delgado, ??? General Assembly 2017, Stockholm, Sweden Aug 13, 2017
Outline Motivation: Improve Transit Agency/Operator Key Functions Objectives Next Steps
Key Transit Agency/Operator Functions A. Off-Line Functions Service and Operations Planning (SOP) Performance Measurement (PM) B. Real-Time Functions Service and Operations Control and Management (SOCM) Customer Information (CI)
System Monitoring, Analysis, and Prediction Key Functions Off-line Functions Real-time Functions Supply Demand Customer Information (CI) Service Management (SOCM) Service and Operations Planning (SOP) ADCS Performance Measurement (PM) System Monitoring, Analysis, and Prediction ADCS: Automatic Data Collection Systems
Examples of ADCS in Decision Support Passenger Flow and System Capacity Public Transport OD Matrix Estimation Performance Measurement (PM) Real time demand estimation and control Customer information
Commercial Dispatcher Console Synoptic Fleet control & monitoring Commercial Dispatcher Comply with frequency, regularity and indicators Smart dispatcher Console Driver assistance On-route regularity Terminal Dispatcher Opening & closing of driver shifts Non commercial movements Surveying Passenger counting App Demand Per stop & line Load profile Fare evasion supervision Users Real-time updates to users Training Dispatchers & drivers
Motivation Benefits of the User Benefits of the Operator Improvement of travel planning Knowledge of travel times Reduction of waiting anxiety (Peng et al., 2002) cambiar Benefits of the Operator Better strategic planning of the routes and number of buses Option to generate Priority Traffic Light Systems Maintain bus regularity in passing frequency (Ex: control mechanism)
Main Objective Development of predictive models that take advantage of multi-source heterogeneous data set to predict the state of public transport systems for short and medium term horizons.
Multi-source heterogeneous data sets We are considering are: Transactional data from fare card collection systems, automatic toll collection systems, accidents, etc. Vehicle related data such as GPS tracks, accelerometer, gyro, thermometer, images, etc. Publicly available data, such demographics, economics, social media, weather, incidents, etc.
Specific Objectives Identify advantages and disadvantages of different data sources for building predictive models. Provide and generate recommendation to users at individual level based on the status of the system and given their preferences. Provide and generate recommendation to operators to offer a better service to their customers.
Methods We consider the collection of multi-source data set to calibrate and then validate the proposed forecast methods. We are considering the construction of an architecture based on a mixture of modern deep learning models to jointly process multi- source data from transactional, vehicles, and publicly available. Propose optimization methods to take advantage of forecast developed in the project so that better decision can be made.
Next Steps: Improve Level of Service Prediction Travel Speed Crowdedness / Seat Availability etc. Transit Route Recommendation … Looking for partners
General Assembly 2017, Stockholm, Sweden Development of predictive models for public transport operations to improve user information, indicators and real-time forecast for operators and transit agencies motivacion Ricardo Giesen, Juan Carlos Herrera, Hans-Albert Löbel, Felipe Delgado, ??? General Assembly 2017, Stockholm, Sweden Aug 13, 2017