You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@systemml.apache.org by "Niketan Pansare (JIRA)" <ji...@apache.org> on 2016/09/29 19:14:20 UTC
[jira] [Created] (SYSTEMML-990) Add mllearn and scala wrappers for
stepwise GLM
Niketan Pansare created SYSTEMML-990:
----------------------------------------
Summary: Add mllearn and scala wrappers for stepwise GLM
Key: SYSTEMML-990
URL: https://issues.apache.org/jira/browse/SYSTEMML-990
Project: SystemML
Issue Type: Task
Components: APIs
Reporter: Niketan Pansare
See https://apache.github.io/incubator-systemml/algorithms-regression.html#stepwise-generalized-linear-regression for usage.
Since this is a starter task, I describe the steps to complete this task:
1. Implement a scala class (which inherits from BaseSystemMLRegressor) similar to https://github.com/apache/incubator-systemml/blob/master/src/main/scala/org/apache/sysml/api/ml/LinearRegression.scala
2. Modify getTrainingScript and getPredictionScript to specify the parameters used. See the algorithm documentation for these parameters.
3. Ensure that you implement appropriate traits to accept hyperparameters (eg: HasLaplace, HasIcpt, HasRegParam, HasTol, etc). These traits are available at https://github.com/apache/incubator-systemml/blob/master/src/main/scala/org/apache/sysml/api/ml/BaseSystemMLClassifier.scala#L36
4. Implement a python class (that extends BaseSystemMLRegressor) with constructor similar to https://github.com/apache/incubator-systemml/blob/master/src/main/python/systemml/mllearn/estimators.py#L218 which essentially accepts the hyperparameters and invokes the scala side methods (example: self.estimator.setLaplace(laplace))
5. Update the algorithm documentation by specifying the usage as well as examples.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)