Unable to optimize parameters in Fit Trend-SVM model

mmaragmmarag MemberPosts:35Guru
edited November 2018 inHelp

Hello,

i am trying to oprimize the SVM parameter C using GRID optimization in an attempth to fit the trend of a stock market as better as possible. I get the error message: Trend(label) attribute is duplicate. What am i doing wrong?

here is the process:







<运营商激活= " true "类= compati“过程”bility="7.3.000-BETA" expanded="true" name="Process">











































































< portSpacing端口= " source_exampleset" spacing="0"/>






























Thanks

MM

PS: here is the stack trace

Exception: java.lang.IllegalArgumentException
Message: Duplicate attribute name: trend(label)
Stack trace:

com.rapidminer.example.SimpleAttributes.register(SimpleAttributes.java:124)
com.rapidminer.example.SimpleAttributes.add(SimpleAttributes.java:203)
com.rapidminer.example.AbstractAttributes.addRegular(AbstractAttributes.java:94)
com.rapidminer.operator.preprocessing.series.filter.Trend.doWork(Trend.java:153)
com.rapidminer.operator.Operator.execute(Operator.java:1005)
com.rapidminer.operator.execution.SimpleUnitExecutor.execute(SimpleUnitExecutor.java:76)
com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:812)
com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:807)
java.security.AccessController.doPrivileged(Native Method)
com.rapidminer.operator.ExecutionUnit.execute(ExecutionUnit.java:807)
com.rapidminer.operator.meta.ParameterIteratingOperatorChain.executeSubprocess(ParameterIteratingOperatorChain.java:262)
com.rapidminer.operator.meta.ParameterIteratingOperatorChain.getPerformanceVector(ParameterIteratingOperatorChain.java:305)
com.rapidminer.operator.meta.GridSearchParameterOptimizationOperator.computeCurrentPerformance(GridSearchParameterOptimizationOperator.java:113)
com.rapidminer.operator.meta.GridSearchParameterOptimizationOperator.doWork(GridSearchParameterOptimizationOperator.java:177)
com.rapidminer.operator.Operator.execute(Operator.java:1005)
com.rapidminer.operator.execution.SimpleUnitExecutor.execute(SimpleUnitExecutor.java:76)
com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:812)
com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:807)
java.security.AccessController.doPrivileged(Native Method)
com.rapidminer.operator.ExecutionUnit.execute(ExecutionUnit.java:807)
com.rapidminer.operator.OperatorChain.doWork(OperatorChain.java:428)
com.rapidminer.operator.Operator.execute(Operator.java:1005)
com.rapidminer.Process.run(Process.java:1195)
com.rapidminer.Process.run(Process.java:1091)
com.rapidminer.Process.run(Process.java:1044)
com.rapidminer.Process.run(Process.java:1039)
com.rapidminer.Process.run(Process.java:1029)
com.rapidminer.gui.ProcessThread.run(ProcessThread.java:65)

Book1.zip 125.2K
Tagged:

Answers

  • Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:1,761Unicorn

    @stevefarrplease move this thread to the Studio forum. This sould not be in the Building Block forum.

    Thanks.

  • Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:1,761Unicorn

    Hi,

    I checked out your process and see a few things that are not quite correct. You vary the C by decimal increments: 1.1, 1.2, 1.3, that's not very effecient and you won't get anything bump in performance. May I suggest using ten steps and varying from 0 to 10,000?

    In addition, with time series you have to use the Sliding Window Validation and nest it with the Forecast Performance operator, then you can optimize trend accuracy.

    Update: See XML below as an example.







    <运营商激活= " true "类= compati“过程”bility="7.2.003" expanded="true" name="Process">

































    <连接from_op = from_por应用验证模型t="labelled data" to_op="Measure Forecast Perf" to_port="labelled data"/>


































Sign InorRegisterto comment.