Intro to ML

Our APIs have Machine Learning (ML) capabilities baked into them. If you’re using our Building Blocks, you’re collecting data in a way that is already plugged into our ML tool set. These docs will explain what you can do with our ML tools and how to do it.

We have two main types of ML models, recommendations and similarities. We’re continuously working on expanding that list.

Once you’re plugged into our APIs, you have access to several out-of-the-box recommenders and similarity models that apply to the most common use cases.

For much more flexibility, check out the Advanced section.

All ML models described in these docs are different, but share a set of common features and parameters. You can either use the model_name to get one of our Out-of-the-Box models or follow the Advanced section to customize a model to your needs.

In general, you will go through the following three phases for all ML models:

  1. Create the model.

  2. Train the model.

    1. The model can be retrained later when more data is collected.

    2. An automatic evaluation that shows the validity of the model is created while the model is trained. This indicates how predictive the results should be.

  3. Query the recommendations or predictions online.