By: Kelly Haberl and Hannah Dao

Lyft’s Fleet business started in 2016 with the mission of powering the world’s best transportation through fleet services that our customers can trust and depend on. Fleet supports Lyft’s drivers and riders by providing access to rental vehicles for drivers to earn both on and off of Lyft’s rideshare platform and by offering trusted and convenient car rental services for all Lyft users. As part of Lyft’s long-term vision, our Fleet team will lead Lyft’s efforts in a future built on electric vehicles and eventually autonomous vehicles.

Fleet Science works with five key Fleet businesses including Express Drive, Lyft Rentals, Vehicle Services, Fleet Platform, and Lyft Autonomous.

• Express Drive: Express Drive provides gig rentals to drivers and is building the capability to manage large fleets of vehicles for businesses, including buying and selling vehicles. We currently provide weekly vehicle rentals in 34 regions across the country, offering a variety of vehicle types including hybrid and electric vehicles.
• Lyft Rentals: Lyft Rentals provides vehicles for riders who are looking to rent a vehicle for personal use such as a weekend vacation. These rentals are currently offered in 28 regions nationally.
• Vehicle Services: Lyft Vehicles Services supports our drivers by offering high-quality, trusted vehicle service through end-to-end capabilities such as maintenance, roadside assistance, and collision services.
• Fleet Platform: Fleet Platform focuses on building a best-in-class suite of Fleet solutions, unifying customer and vehicle journeys with capabilities such as fleet planning and allocation, vehicle quality control, and operational workflows.
• Lyft Autonomous: Lyft Autonomous is building the technology and partnerships to deliver autonomous rideshare experiences, enabling AV partners to plug into Lyft’s network.

What is Fleet Data Science?

On Fleet Data Science, we work on a variety of data-driven analyses from shaping critical business decisions to building algorithms that power our products (more to come on these projects later!). Fleet is a fast growing business within Lyft, so our work often mimics this agile, fast-paced style as we help shape the future of Fleet at Lyft.

As Lyft’s Fleet vision grows, our team has rapidly expanded as well. Our Fleet Data Science team currently consists of 10 scientists, spanning both of the Data Science roles at Lyft (for more information on these roles, check out this post):

• Data Scientist, Decisions: Utilizing a deep understanding of the business to develop decision frameworks that drive alignment on the most impactful problems and solutions.
• Data Scientist, Algorithms: Develop models that power internal and external production systems, applying a combination of optimization, machine learning, and inference methods.

Our team is structured into pods, allowing each scientist to take ownership for a part of the Fleet business. A few examples of these pods include:

• Marketplace: understanding supply and demand in the rental marketplace to ensure effective fleet management and growth
• Renter Experience: ensuring affordable, fair, and delightful in-rental experiences
• Pick Up & Drop Off: creating a seamless process for vehicle pick up and drop off at our rental locations

What do Fleet Data Scientists work on?

Below we outline several examples of projects our Fleet Data Scientists work on:

Rental Pricing

Data Scientists play a key role in how Lyft sets rental rates and understands the relationship between prices and rental demand. Examples of questions we analyze include:

1. What pricing structure is best for renters? Should we charge a flat weekly / daily rate or charge a small fee per mile driven? What are the impacts of each pricing structure on renters and on Lyft?
2. What is the impact to rental demand if a discount on a rental rate is offered? Should this discount be a dollar discount or other perks (e.g., free mileage)? How do we decide when and where to deploy discounts?

A key tool used to answer these questions is A/B testing, as our team has deployed a variety of experiments (user-split, region-split, etc.) and advanced causal inference to understand the impact of pricing, recommend solutions to our cross-functional partners, and drive product changes.

Fleet Forecasting

In order to efficiently manage our fleet, it is important for us to create an intelligent forecast of the market supply (vehicle availability) and demand (drivers wanting to rent). On this highly collaborative workstream, we work closely with our cross-functional stakeholders such as the Operations team to develop models which generate the supply and demand forecast weekly at a regional level.

Due to the work of our Data Scientists, our forecast system evolved from a manual process to an automated system which reduces operational effort and time for the team. Moreover, the automated forecasting system enables various important features for our fleet such as fleet allocation optimization, which helps ensure we have rental vehicles in the right location for our renters. In terms of technical approach, we started with simple statistical models but soon improved our algorithms with advanced machine learning techniques such as times series forecasting and classification models.

Renter Experience

Working to ensure a positive rental experience, this team provides data-driven recommendations on questions such as:

1. How are renters paying for their vehicle? Should we charge Express Drive renters on a weekly cadence or monthly? How are our renters driving on the Lyft rideshare platform and how are we enabling them to earn on other platforms?
2. What are the main pain points for our renters? What changes can we make to address these areas?
3. As we build a fleet focused on electric vehicles, how will renters charge their vehicles during a rental? What cost benefits will our electric vehicles provide to our renters?

Fleet Management

One of our goals on the Lyft Fleet team is to power large-scale fleets of traditional and autonomous vehicles, micromobility assets, and high-capacity vehicles for organizations and cities across the world. Our Data Science team plays an important role in this vision by leading our experimentation efforts and providing data insights to develop product features for our fleet management product. Our Data Science team is working to understand:

1. How do we streamline the vehicle lifecycle to minimize downtime and maximize rental capabilities?
2. How do we optimize fleet allocations across regions based on changing rental demand, vehicle costs, and local team support?
3. Can we automate vehicle assignment for new renters to minimize manual efforts?

Are you as excited as we are? Lyft Data Science is hiring! We’re always looking for great additions to our growing Data Science team. Join us!