Overview
Fire Insights is now integrated with Amazon Sagemaker.
The workflows can now use Sagemaker Processors in Fire to run Machine Learning Model building on Sagemaker.
Details on using Sagemaker with Fire Insights is available here:
http://docs.sparkflows.io/en/latest/aws/sagemaker.html
Sagemaker Workflow
Below is a workflow in Fire which connects to Sagamaker. It does the following:
Reads in a libsvm file. You can also read in a data file, create features and use the VectorAssembler processor to create the features column.
Run XGBoost Estimator on Sagemaker.
Reads in another libsvm file.
Uses the Prediciton Processor do use the model generated by Sagemaker to make predictions.
Finally prints the predictions. The prediction can also be save into files etc. by using the Save Processors.
XGBoost Configuration
Below is the configuration set of the XGBoost Processor.
Workflow Execution
Below are the results of executing the Workflow.