We integrate ml-engine API model and version functions to Drupal. Video. Refer to the v1.projects.models section of the page for job method details.

Comments

boaloysius created an issue. See original summary.

boaloysius’s picture

boaloysius CreditAttribution: boaloysius as a volunteer and at Google Summer of Code commented
Issue summary: View changes
boaloysius’s picture

boaloysius CreditAttribution: boaloysius as a volunteer and at Google Summer of Code commented

This week was kick-started by working on the basic architecture of the module. We have created six Drupal services and an MLEngineBase class. As any request to the Google cloud server should be for a service like CloudNaturalLanguage, CloudSpeechAPI, Compute, Freebase, YouTube etc, we have created a Drupal service to provide CloudMachineLearningEngineService for every request we make to the cloud server. We have created APIs for Training Job, Models and its Version. They have create, get, list, and cancel functions. They are designed to give a similar format for the response. array( ['success]=> 1, ['response']=>$response ) and  array( ['success]=> 0, ['response']=> json_decode(($ex->getMessage(), true)['error'])). They work with forms to send requests to the server. These services extend MLEngineBase class which provides global variables and functions like project name and create_service(). We have Prediction and Project services that verify the project credential and make predictions. Please check the commit 3564195 to see the major design improvement. 

Beside this, we have some minor design changes. I feel it easy to explain the progress through a video than text. 

Youtube link