You are viewing the RapidMiner Studio documentation for version 8.0 - Check here for latest version

Collect (RapidMiner Studio Core)

Synopsis

This operator combines multiple input objects into a single collection.

Description

The Collect operator combines a variable number of input objects into a single collection. It is important to know that all input objects should be of the same IOObject class. In the Process View, collections are indicated by double lines. If the input objects are collections themselves then the output of this operator would be a collection of collections. However if the unfold parameter is set to true then the output will be the union of all elements of the input collections. After combining objects into a collection, the Loop Collection operator can be used to iterate over this collection. The Select operator can be used to retrieve the required element of the collection.

Collections can be useful when you want to apply the same operations on a number of objects. The Collect operator will allow you to collect the required objects into a single collection, the Loop Collection operator will allow you to iterate over all collections and finally you can separate the input objects from collection by individually selecting the required element by using the Select operator.

Input

  • input (IOObject)

    This operator can have multiple inputs. When one input is connected, another input port becomes available which is ready to accept another input (if any). The order of inputs remains the same. The object supplied at the first input port of the Collect operator becomes the first element of the resultant collection. It is important to note that all input objects should be of the same IOObject class.

Output

  • collection (Collection)

    All the input objects are combined into a single collection and the resultant collection is delivered through this port.

Parameters

  • unfoldThis parameter is only applicable when the input objects are collections. This parameter specifies whether collections received at the input ports should be unfolded. If the input objects are collections themselves and the unfold parameter is set to false, then the output of this operator would be a collection of collections. However if the unfold parameter is set to true then the output will be the union of all elements of the input collections. Range: boolean

Tutorial Processes

Introduction to collections

This Example Process explains a number of important ideas related to collections. This Example Process shows how objects can be collected into a collection, then some preprocessing is applied on the collection and finally individual elements of the collection are separated as required.

The 'Golf' and 'Golf-Testset' data sets are loaded using the Retrieve operator. Both ExampleSets are provided as inputs to the Subprocess operator. The subprocess performs some preprocessing on the ExampleSets and then returns them through its output ports. The first output port returns the preprocessed 'Golf' data set which is then used as training set for the Decision Tree operator. The second output port delivers the preprocessed 'Golf-Testset' data set which is used as testing set for the Apply Model operator which applies the Decision Tree model. The performance of this model is measured and it is connected to the results port. The training and testing ExampleSets can also be seen in the Results Workspace.

Now have a look at the subprocess of the Subprocess operator. First of all, the Collect operator combines the two ExampleSets into a single collection. Note the double line output of the Collect operator which indicates that the result is a collection. Then the Loop Collection operator is applied on the collection. The Loop Collection operator iterates over the elements of the collection and performs some preprocessing (renaming an attribute in this case) on them. You can see in the subprocess of the Loop Collection operator that the Rename operator is used for changing the name of the Temperature attribute to 'New Temperature'. It is important to note that this renaming is performed on both ExampleSets of the collection. The resultant collection is supplied to the Multiply operator which generates two copies of the collection. The first copy is used by the Select operator (with index parameter = 1) to select the first element of collection i.e. the preprocessed 'Golf' data set. The second copy is used by the second Select operator (with index parameter = 2) to select the second element of the collection i.e. the preprocessed 'Golf-Testset' data set.