RapidMiner6操作员参考手册

3星 · 超过75%的资源需积分: 50 184 浏览量更新于2024-07-21 收藏 7.5MB PDF 举报

"RapidMiner6操作员参考手册提供了关于该数据挖掘工具的详细信息，包括各种操作、参数设置、循环、分支等组件的使用方法。" 在RapidMiner这一强大的数据挖掘和机器学习平台上，用户手册是理解其功能和提升工作效率的关键。RapidMiner6的操作员参考手册涵盖了多个重要知识点： 1. **过程控制**：这部分介绍了如何处理异常（HandleException）、合并路径（JoinPaths）和乘法操作（Multiply）。例如，HandleException用于捕获和处理执行过程中的错误，而JoinPaths则允许合并多个数据流。 2. **发布与召回**：PublishtoApp和Recall.fromApp这两个操作分别用于将工作流发布到RapidMiner Studio之外的应用，并从这些应用召回工作流程，实现了工作流的共享和持续运行。 3. **记忆与调度**：Remember操作可以保存中间结果，以便后续使用；ScheduleProcess则允许用户设置定时任务，使工作流按照预设时间自动运行。 4. **参数管理**：包括CloneParameters、OptimizeParameters（进化式、网格搜索、二次优化）、SetParameters等。这些操作用于复制参数、优化模型参数以提升模型性能，以及直接设置参数值。 5. **循环机制**：RapidMiner支持多种类型的循环，如Loop、LoopAttributeSubsets、LoopAttributes等，可以对数据集的不同部分、属性、批次、集群、数据集、示例、文件、标签、参数和值进行迭代。例如，LoopandAverage和LoopandDeliverBest分别用于计算循环内的平均值和选取最佳结果。 6. **分支结构**：Branch操作允许根据特定条件执行不同的子流程，SelectSubprocess则用于选择和运行预先定义的子流程，增强了流程的灵活性。手册还详细列出了各个操作的参数设置，如1.1节中的CloneParameters、OptimizeParameters等，以及1.2节中的各种循环操作，和1.3节的Branch和SelectSubprocess。这些参数和操作的详细描述为用户提供了丰富的工具来构建和定制复杂的数据挖掘和分析流程。通过深入理解和熟练运用RapidMiner6操作员参考手册中的知识点，用户能够高效地构建、调试和优化数据挖掘模型，实现数据的预处理、特征工程、建模、验证和部署等一系列工作，从而提升数据分析的质量和效率。

1. Process Control

Parameters

exception macro (string) This parameter speciﬁes the name of the macro whichwillstore the

error message (if any). This macro can be accessed by other operators by using ‘%{macro-

_name}’ syntax.

Tutorial Processes

Using Try/Catch for exception handling

Main Process

Iris

out

Handle Exc...

out

inp res

res

Figure 1.1: Tutorial process ‘Using Try/Catch for exception handling’.

The goal of this Example Process is to deliver the ‘Iris’ data set after renaming its attributes. In

case there is an error, the original ‘Iris’ data set should be delivered along with the error message.

The ‘Iris’ data set is loaded using the Retrieve operator. A breakpoint is inserted here so that

you can have a look at the ExampleSet. You can see that there are four regular attributes a1, a2,

a3 and a4. The Handle Exception operator is applied next. Have a look at the subprocesses of this

operator. The Rename operator is applied in the Try subprocess for renaming the attributes. The

old name and new name parameters of the Rename operator are deliberately left empty which

is an error because these are mandatory parameters. The Catch subprocess takes the original

ExampleSet and applies the Log operator and delivers the ExampleSet without modiﬁcations to

the output. The Log operator is used for recording the error message in case there is an error.

Execute the process and switch to the Results Workspace. You can see that the attributes of

the ExampleSet have not been renamed. This is because there was an error in the Try subprocess,

therefore it was not executed. The error message can be seen in the log which says that values for

mandatory parameters of the Rename operator were not provided. The original ExampleSet can

be seen in the Results Workspace because when the Handle Exception operator encountered an

error in the Try subprocess, it stopped its execution and executed the Catch subprocess instead

and delivered its results.

Now set the old name and new name parameters of the Rename operator to ‘label’ and ‘new-

_label’ respectively and run the process again. This time you can see that the attributes have

been renamed and there is no error message. This is so because the Handle Exception operator

ﬁrst tried to execute the Try subprocess. Since there was no error in it, its results were delivered

and the Catch subprocess was not executed.

Tutorial Processes

Multiplying data sets

Main Process

Labor-Nego...

out

Multiply

inp out

out

Generate ID

exa exa

ori

Select Attri...

exa exa

ori

Replace Mis...

exa exa

ori

pre

inp res

res

Figure 1.3: Tutorial process ‘Multiplying data sets’.

In this Example Process the Retrieve operator is used to load the ‘Labor-Negotiations’ data

set. A breakpoint is inserted after this operator so that the data can be viewed before applying

the Multiply operator. You can see that this data set has many missing values. Press the green-

colored Run button to continue the process.

4 copies of the data set are generated using the Multiply operator. The Replace Missing Values

operator is applied on the ﬁrst copy. The Select Attributesoperator is applied on the second copy.

The Generate ID operator is applied on the third copy and the forth copy is connected directly

to the results port.

The Replace Missing Values operator replaces all the missing values with the average value of

that attribute. As this is a change in data instead of a change in attributes, this change is made

in all the copies.

The Select Attributes operator selects one attribute i.e. the duration attribute. Please note

that special attributes are also selected even if they are not explicitly speciﬁed. Thus the special

attribute of the Labor-Negotiations data set (i.e. the Class attribute) is automatically selected.

Results from this operator show only two attributes: the Duration attribute and a special at-

tribute (the Class attribute). As this is a change in attributes instead of a change in data, it is

only applied to this copy and all other copies are not affected. Similarly the Generate ID opera-

tor adds a new attribute (the id attribute) to the data set. As this is not a change in data (it is a

change in attributes), it is relevant only for this copy and other copies are not affected.

The Last copy generated by the Multiply operator is connected directly to the results port with-

out applying any operator. This copy is not the same as the input ExampleSet. This copy has

no missing values. This is because of the Replace Missing Values operator, it made a change in

data and changes in data are reﬂected in all copies.

剩余919页未读，继续阅读

AI_盲

粉丝: 528
资源: 6

RapidMiner6操作员参考手册

RapidMiner数据分析与挖掘实战——中文

RapidMiner6使用手册

数据挖掘工具Rapidminer基础介绍

RapidMiner5.0使用手册

rapidminer使用手册 [RapidMiner数据分析与挖掘实战] 全17章

Rapidminer中文手册

Rapidminer开发者版

rapidminer使用手册

rapidminer中文手册

rapidminer-examples:使用RapidMiner的示例流程和POC

最新资源