Emma Rosenberg

Machine learning in e-commerce

A research project to consider how we could improve user experience on e-commerce products using machine learning techniques, with the aim of giving other team members and stakeholders an overview and recommendations for future work.


Process

To explore how machine learning could be utilised on Farfetch, we needed to understand how it was already being employed on site, and consider whether features using this technology were solving user problems, as well as investigating possible new uses for the technology. I'm pretty interested in this topic in general, and had already spent some of my free time researching the topic, including taking this Coursera course over a few weekends to get a basic introduction to Python and an overview of some machine learning techniques.


We collaborated with data science team members to get an overview of how the technology was being used already on site. Then, in a team of four, we workshopped ideas and sketched and protoyped solutions, gathering feedback along the way from co-workers and via some speedy user testing to tentatively validate some of the ideas.


↑ invision prototype, personalised content



Outcomes and next steps

The outcome was a series of proposed solutions, including using visual similarity algorithms to help users find similar products to ones they're interested in while browsing, and using chat platforms to send recommendations directly to users in a format that's familiar and easy to use. I prepared a presentation deck of the project work and the results were communicated back to key stakeholders and members of the teams. Click here to see the presentation.


↑ invision prototype, chat to send recommended products