This is a sample project that shows the basic features of the bigmlcom/bigml-java bindings.
It is a Eclipse project with a simple class
org.bigml.sample.BigMLSampleClient
that shows how to create resources and
local predictions using these bindings.
The examples include:
A basic predictive workflow using BigML's API:
- Create a Source
- Create a Dataset
- Create a Model
- Create a Prediction
Creatinge local predictions by downloading the contents
of the remote Model
JSON object.
- Create a LocalPredictiveModel
- Create a prediction using the
LocalPredictiveModel
instance
Evaluating your model by sampling your existing dataset:
- Create a training sampled Model
- Create an Evaluation with the out of bag test data
Creating some unsupervised models:
The workflow needed to create Topic Distributions:
- Create a Topic Model
- Create a Topic Distribution
How to change the properties of the fields in your
Source
or Dataset
.
How to create a WhizzML Script
and execute
it for some inputs
by creating an Execution
.
You'll need to set your own API credentials in
src/main/resources/binding.properties
or in the environment
variables BIGML_USERNAME
and BIGML_API_KEY
respectively.
The instance of the BigMLClient
will use these credentials to authenticate
your calls to the API.
The binding.properties file should contain the constants below, that define the domain that the calls will point to, and the user authentication information
BIGML_URL=https://bigml.io/andromeda/
BIGML_USERNAME=
BIGML_API_KEY=
If you use environemnt variables you can add the following lines to your
.bashrc
or .bash_profile
to set those variables automatically when you
log in:
export BIGML_API_KEY=ae579e7e53fb9abd646a6ff8aa99d4afe83ac291
You can find your API Key at https://bigml.com/account/apikey