<aside> <img src="/icons/info-alternate_blue.svg" alt="/icons/info-alternate_blue.svg" width="40px" />

Intro

EcoScore and Coaching is a driver and trip scoring engine, part of the eco-fleet (fleet management solution) built for Paris Olympics 2024, with a goal to help achieve a fleet average of <40g/km CO2 emissions (halve of the previous olympics) across the Toyota vehicle fleet used at the Olympics.

</aside>

<aside> <img src="/icons/condense_yellow.svg" alt="/icons/condense_yellow.svg" width="40px" />

Solution

Scoring engine uses a regression model (ElasticNet) to score trips and drivers between 0 and 100 on energy efficient driving behaviours, and a more detained breakdown of the behaviours using sub-scores. Additional it generates recommendation for drivers on how to improve their energy efficiency using Explainable AI that uses model-agnostic interpretability methods to generate relevant and robust recommendations.

ML Observability and Monitoring tools implemented using Evidently AI, Gitlab pages, and DataDog to monitor and alert on data drift, data quality, regression performance, advice (explainability) drift, bias and fairness.

</aside>

<aside> <img src="/icons/bullseye_green.svg" alt="/icons/bullseye_green.svg" width="40px" />

Project Outcome

Within Paris Olympics (and Para Olympics) 2024, the solution led to 500k+ of total trips across 2.5k+ vehicles being scored on energy efficiency and each trip explained on what behavioural change would impact energy efficiency the most → ultimately helping achieve the goal of <40g/km emissions.

The EcoScore product has further gained interest in other fleet management solutions, and is serving in production across multiple fleet products.

</aside>

<aside> <img src="/icons/archery_purple.svg" alt="/icons/archery_purple.svg" width="40px" />

Soft Skills Demonstrated or Developed

• Stakeholder Management and Communication: Understanding the customer requirements, taking the customer on a journey to use machine learning for the given purpose breaking down complex data science concepts into simple business and customer friendly stories. Aligning the product objectives to customer expectations. Served as the main representative of DS on the project liaising with Product, Privacy and Engineering.
• Technical Ownership & Leadership: Led the entire project end-to-end, from discovery, to model development and evaluation, to deployment and continuous monitoring and ongoing product support. </aside>