ML architecture using newest tools: Predicting near-future passenger demand for Hyperloop
Who is this presentation for?Data scientists or analysts
Hyperloop is a new mode of high-speed mass transportation not only able to reduce transit times but also afford passengers an on-demand experience. In contrast to traditional mass transportation systems that update fixed operating schedules year by year, the Virgin Hyperloop Mass Transportation System (VHOMTS) can dynamically redistribute vehicles in real time to minimize the passenger wait time even in the highest demand peaks and reduce its service expenditures in downtimes without impacting passenger experience. To do so efficiently, the VHOMTS uses prediction of passenger origin-destination demand in the next hours.
Patryk Oleniuk and Sandhya Raghava show the process for the creation of Virgin Hyperloop One’s passenger origin-destination demand prediction model for driving simulation and analysis of system performance under a wide range of operating conditions and real-time vehicle redistribution.
They explore historical data gathering, data transformations, and pattern analysis, as well as model selection and training using recent DNNs and compare it with other conventional methods. You’ll understand the architecture and methodology the company used to train on GBs of demand data, perform distributed training of thousands of models, sweep multiple model parameters in parallel, verify, and save them for future use. And you’ll cover selected topics from feature engineering, such as the impact of weather and surrounding events on the prediction accuracy.
To substantiate the impact of the demand prediction, they present a set of case studies for prospective Hyperloop markets. The advantages of a forecast-based versus purely reactive scheduling system are quantified in terms of passenger wait times and system use using simulated operations. Patryk and Sandhya’s methodology for passenger origin-destination demand modeling could also be applied to other modes of transportation, like cars, scooters, or cargo.
- A basic knowledge of machine and deep learning
What you'll learn
- Discover Virgin Hyperloop One data platform and data sources
- Understand data preprocessing with Koalas (pandas for Spark), demand prediction with Keras and performance comparison with ARIMA methods, and experiment tracking with MLflow
Virgin Hyperloop One
Patryk Oleniuk is a data engineer at Virgin Hyperloop One, a company building the fifth mode of transportation. Previously, he was at CERN, where he wrote test software for the world’s biggest particle accelerator, National Instruments, and Samsung R&D. He graduated from EPFL (Swiss Federal Polytechnique in Lausanne) with an information technologies major. When he isn’t glued to a computer screen, he spends time road-tripping California with his friends.
Virgin Hyperloop One
Sandhya Raghavan is a senior data engineer at Virgin Hyperloop One, where she helps building the data analytics platform for the organization. She has 13 years of experience working with leading organizations to build scalable data architectures, integrating relational, and big data technologies. She also has experience implementing large-scale, distributed machine learning algorithms. Sandhya holds a bachelor’s degree in computer science from Anna University, India. When Sandhya isn’t building data pipelines, you can find her traveling the world with her family or pedaling a bike.
Leave a Comment or Question
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
Premier Diamond Sponsors
Premier Exhibitor Plus
For conference registration information and customer service
For more information on community discounts and trade opportunities with O’Reilly conferences
For information on exhibiting or sponsoring a conference
For media/analyst press inquires