Skip to content

Latest commit

 

History

History
85 lines (69 loc) · 2.86 KB

guide.md

File metadata and controls

85 lines (69 loc) · 2.86 KB
Raw

Bindings info for developers

Welcome to the documentation site for BigML's bindings developers. We'll try to provide the latest news and issues affecting the bindings for the BigML REST API and associated resources.

Each binding has its own particular structure, but generally speaking, these are the function levels they provide:

Basic functions:

  • Connection client: manages the connection to BigML's servers and stores user credentials.

  • REST API interface: manages the REST calls to BigML's API for all the available resources, namely:

    • Sources
    • Datasets
    • Models
    • Predictions
    • Evaluations
    • Batch Predictions
    • Ensembles
    • Clusters
    • Centroids
    • Batch Centroids
    • Anomaly Detectors
    • Anomaly Scores
    • Batch Anomaly Scores
    • Projects
    • Samples
    • Correlations
    • Statistical Tests
    • Logistic Regressions
    • Associations
    • Topic Models
    • Topic Distributions
    • Time series
    • Forecasts
    • Deepnets

    You can learn about the properties of these resources and the API calls' arguments in the developers section of our website.

Advanced functions:

  • Advanced methods: BigML offers the ability to download datasets or batch predictions in a csv format. Some bindings provide methods to download such files.

  • Local models, ensembles, clusters, anomaly detectors, etc.: BigML's predictive models can be downloaded as JSON structures. Bindings can provide objects built on this JSON which are able to produce local predictions, centroids or anomaly scores.

  • Fields utilities: Bindings can provide utilities that help listing, or extracting specfic field information from the corresponding attribute in each resource's JSON.

The Python, Node.js, Java, C#, iOS, Clojure bindings are maintained by BigML's team and are usually. Usually, Python and Node.js are updated to offer the latest added features. They can be used as reference in new bindings development.

Documentation and tests:

To the best interest of the bindings' users and the entire community, we highly encourage our bindings' authors to document and test all new functions as they are added. Code examples and use cases are especially useful to speed learning, and good quality code needs good quality documents and tests to reach its goal.

Backend new features:

There's an API news notice in this report that will be updated to show the features recently added in the API. It will also provide any information needed to create the corresponding bindings interface. We will keep a historic record for the latest new features as they are updated. We keep a record of the development level of each binding in a shared spreadsheet.