Screenshot
POC

Freitag product map

A map of products where products that "look" the same are close to each other. This is based on deeplearning and t-sne clustering.

Description

What is it?

This visualization arranges images of Freitag bags in a 2D grid according to their similarities (the more similar, the closest).

Demo

See screenshot. It displays the bags arranged in a "similarity-grid".

Benefit

  • For users: Each Freitag bag being a unique piece, this visual map could be an innovative way of browsing through products. Instead of starting from a specific type of bag, customers start in an area with design features that they like (e.g. colors or patterns). Items in the same style are listed close by, rather than items with similar function as in a classical web shop navigation.
  • For Freitag: Better online sales thanks to improved user experience and easier discovery of products across categories.

Estimation

  • About 1 person day

Data

  • Photos of roughly 1000 Freitag bags

Technologies used

  • TSNE clustering, deep learning

Further vision

With more time, you could create a JS map to browse products like you browse a map, and then zoom and click into those.
With more data, not just bags, but also complementary accessories could be added.