DEPRECATION WARNING
This code example was intended for use by the legacy Skafos platform and is no longer being maintained. On 05/29/2019, a new version of Skafos was released, streamlining model delivery to the edge.
Sign-up for an account, join our Slack community, and explore some example models to get started.
The following repo contains example code for training a product recommendation model on Skafos based on the Recommendation Engine Quickstart model. For this example, we used data from Thingiverse: a community where users interested in the world of 3D printing technology share different 3D models they have created. We built and deployed a model on Skafos to generate personalized recommendations of 3D models to users.
The components of this repo are:
product_recommender.ipynb- a Python notebook that trains and saves a product recommendation model. It walks through the steps of data collection, model training, and deployment.utilities/- a directory that contains helper functions used byproduct_recommender.ipynb.requirements.txt- a file describing all required Python dependencies.
- The product recommender is trained on open Thingiverse data retrieved with their REST API. Users can "like" different 3D model designs. This interaction data was used for model training. Here is a snippet of what this data looked like:
| userId | thingId |
|---|---|
| 1 | 343455 |
| 1 | 345353 |
| 2 | 643363 |
- The model can generate personalized recommendations to a user based on which 3D model designs they like and ones that others have liked.
- In order to get data from the Thingiverse API, you must register a developer account with them.
- Thingiverse API Python wrapper ~ adapted from: https://github.com/hroncok/thingiverse