Introduction 

Briefly write the need for this feature

Goals

Clearly state the design goals/requirements for this feature 

User Stories 

• Breakdown of User-Stories 
• User Story #1
• User Story #2
• User Story #3

Design

Consider the following pipeline for the design purpose:

																			TRUE
File (Source) -> CSV Parser(Transform) -> Filter (Transform) -> Condition1--------> Logistic Regression (Sink) 
																	|		           
															FALSE	| 				   TRUE
																	|-----> Condition2-------> Random Forest (Sink)
																				|
																		FALSE	|
																				|		TRUE
																			Condition3--------> Decision Tree (Sink) 

In the above pipeline, we want to execute the classification algorithm based on the runtime argument 'input.algorithm'. We also do not want to run the expensive model generation process if the Filter transform did not produce the records enough to proceed further.

The pipeline is configured with 3 condition nodes:

1. Condition1: output.filter Greater Than 1000 AND input.algorithm Equals 'Logistic Regression'
2. Condition2: output.filter Greater Than 1000 AND input.algorithm Equals 'Random Forest'
3. Condition3: output.filter Greater Than 1000 AND input.algorithm Equals 'Decision Tree'

Representation of the Condition in the Pipeline config

Following is one possible representation of the condition stage in the config json. Since conditions are individual stages, they will also appear in the connections section similar to other stages.

{  
   "name":"Condition1",
   "plugin":{  
      "name":"Condition",
      "type":"condition",
      "label":"Condition1",
      "artifact":{  
         "name":"condition-plugins",
         "version":"1.7.0",
         "scope":"SYSTEM"
      },
      "properties":{  
         "conditions":{  
            "cond1":{  
               "subject":"output.filter",
               "operator":"Greater Than",
               "target":"1000"
            },
            "cond2":{  
               "subject":"input.algorithm",
               "operator":"Equals",
               "target":"Logistic Regression"
            },
            "expressions":[  
               {  
                  "operator":"AND",
                  "operand1operands":["cond1",
                  "operand2":"cond2"]
               }
            ],
            "connectors":{  
               "TRUE":"LogisticRegressionStage",
               "FALSE":"Condition2"
            }
         }
      }
   }
}
      
    

Approach

Approach #1

Approach #2

API changes

New Programmatic APIs

New Java APIs introduced (both user facing and internal)

Deprecated Programmatic APIs

New REST APIs

Deprecated REST API

CLI Impact or Changes

• Impact #1
• Impact #2
• Impact #3

UI Impact or Changes

• Impact #1
• Impact #2
• Impact #3

Security Impact 

What's the impact on Authorization and how does the design take care of this aspect

Impact on Infrastructure Outages 

System behavior (if applicable - document impact on downstream [ YARN, HBase etc ] component failures) and how does the design take care of these aspect

Test Scenarios

Releases

Release X.Y.Z

Release X.Y.Z

Related Work

• Work #1
• Work #2
• Work #3

Future work