Tangram for Elixir

The Tangram Elixir library makes it easy to make predictions with your Tangram machine learning model from Elixir.

Usage

# mix.exs
def deps do
  [
    {:tangram, "~> 0.1.0"}
  ]
end
model = Tangram.load_model_from_file("heart_disease.tangram")

input = %{
  "age" => 63.0,
  "gender" => "male",
  # ...
}

output = Tangram.predict(model, input)

For more information, read the docs.

Platform Support

Tangram for Elixir is currently supported on Linux, macOS, and Windows with AMD64 CPUs. Are you interested in another platform? Open an issue or send us an email at help@tangram.xyz.

Examples

This folder contains two examples, examples/basic.exs and examples/advanced.exs.

Basic

This example demonstrates loading a model from a .tangram file and making a prediction.

To run the example:

$ mix run examples/basic.exs

Advanced

This example demonstrates logging predictions and true values to the Tangram reporting and monitoring web app. Before running the example, run tangram app to start the app running locally, open http://localhost:8080 in your browser, and upload the file examples/heart_disease.tangram to it.

Then you can run the example:

$ TANGRAM_URL=http://localhost:8080 mix run examples/advanced.exs

Now if you refresh the production stats or production metrics tabs for the model you uploaded, you should see predictions and true values being reported.

For more information on reporting and monitoring, read the docs.

Notes

  • On Alpine Linux, Tangram for Elixir requires the gcompat and libgcc libraries to be installed. You can run apk add gcompat libgcc to install them.