the stats were obtained with #datascience analysis of over 1 million shopify stores across the globe.
as part of the project we also built a content based #recommender for recommending similar stores to a given store, based on using #embeddings for product names sold by the store.
for generating similar stores, approximate nearest neighbors #ANN method was used as the vectors representing the stores were 300-dimensional.
@threadreaderapp unroll
@threadreaderapp unroll
• • •
Missing some Tweet in this thread? You can try to
force a refresh
