Decision Tree in ODM

The sum of number of observations from each leaf node is less than the total number of observations in my decision tree model. Am I missing something here?
Thanks,
Avinash

Some rows may not reach the leaves. If there is a NULL value for the splitter attribute in a row, then the row can stop in a branch node. When predicting such a row, we will try the surrogate (if available), but if we cannot use the surrogate, then the scoring process will stop and provide the branch node as the final, predicted node. When building, the row may not follow the surrogate path as it is not available at that phase of build.

Similar Messages

  • Decision Trees with ODM 10.1

    Hi all,
    Does anybody know how to use DTs with ODM R1?. Maybe with the ABN model?.
    Any hint or recommendation to start with this?
    Thanks in advance!

    In ODM 10.1 you can get a type of decision tree using ODM in single feature mode. Unlike a traditional decision tree algorithm (e.g., CART), this type of ABM model creates a tree that, for a given level of the tree, it splits all the nodes along the same attribute. This ABN approach has the advantage of being more robust than traditional decision tree models. However, it works better (more accuracy and deeper trees) if the data set has a large number of rows.
    -Marcos

  • ODM Decision Tree result

    I am using Oracle Data Miner with an Oracle 10g Database. I would like to know if there is a tool or a plug-in for ODM to watch the Decisional Tree, resulted of executing the Decision Tree Algorithm, as a tree. Could you help me?
    Thanks in advance!
    .javier

    HI Javier,
    I know I responded to your internal email on this question but I will respond here as well for the OTN folks.
    ODMr 11 will be providing a full decision tree graphical viewer.
    ODMr 10.2 does not have any plugin capability.
    It is quite reasonable to implement a decision tree viewer of your own, the first step would be to locate a diagram tool that meets your needs and then use the ODM api to get the detail data.
    ODM also has partnerships with Inforsense and SPSS(Clementine) that do have dt viewers but I am not sure if they have fully integrated them for ODM yet.
    Thanks, Mark

  • Decision Trees

    Does ODM support Decision Trees.
    If support, which algorithms.
    Thank you.

    We currently do not support classical decision trees. We do support something called Adaptive Bayes Networks that are "similar" in that they can produce "rules" that provide some level of model transparancy. There is a technical white paper on ABN here http://otn.oracle.com/products/bi/pdf/adaptive_bayes_net.pdf

  • Exporting Decision Tree Rules

    I have built the decision tree model in ODM GUI. I would like to export all the decision rules in XLS or PPT from the built model. What is the appropriate way to do this.
    Thanks,
    Avinash

    Hi Avinash,
    Add a Model Details node to the workflow and connect the Build Class node (or the Model node) to the Model Details node.
    Then edit the Model Details node and select the DT details you are interested in.
    In your case, I think that is the Full Tree details.
    You can then right click and select View Data off of the Model Details node and then ctl-a, ctl-c and paste into spreadsheet.
    Thanks, Mark

  • Failed to open model viewer error, for Decision Tree Model

    Oracle 11.0.2
    SQL Developer 3.0.04
    I am trying to follow ODM tutorial here :
    http://www.oracle.com/webfolder/technetwork/tutorials/obe/db/11g/r2/prod/bidw/datamining/ODM11gR2.htm
    After completing initial analysis, I want to investigate the Decision Tree model more closely. As I choose View Models > CLAS_DT_1_1, I get an error message:
    Failed to open model viewer
    Details are:
    java.lang.RuntimeException: not instanceof RoundedTreeNode
         at oracle.dmt.dataminer.mviewer.dtree.model.TreeNode.assignPropertyChangeListener(Unknown Source)
         at oracle.dmt.dataminer.mviewer.dtree.DecisionTreeEditor.setPropertyChangeListener(Unknown Source)
         at oracle.dmt.dataminer.mviewer.dtree.DecisionTreeEditor.open(Unknown Source)
         at oracle.ideimpl.editor.EditorState.openEditor(EditorState.java:283)
         at oracle.ideimpl.editor.EditorState.createEditor(EditorState.java:184)
         at oracle.ideimpl.editor.EditorState.getOrCreateEditor(EditorState.java:95)
         at oracle.ideimpl.editor.SplitPaneState.canSetEditorStatePos(SplitPaneState.java:232)
         at oracle.ideimpl.editor.SplitPaneState.setCurrentEditorStatePos(SplitPaneState.java:195)
         at oracle.ideimpl.editor.TabGroupState.createSplitPaneState(TabGroupState.java:102)
         at oracle.ideimpl.editor.TabGroup.addTabGroupState(TabGroup.java:379)
         at oracle.ideimpl.editor.EditorManagerImpl.createEditor(EditorManagerImpl.java:1403)
         at oracle.ideimpl.editor.EditorManagerImpl.createEditor(EditorManagerImpl.java:1337)
         at oracle.ideimpl.editor.EditorManagerImpl.openEditor(EditorManagerImpl.java:1263)
         at oracle.ide.editor.EditorUtil.openDefaultEditorInFrame(EditorUtil.java:164)
         at oracle.dmt.dataminer.workflow.WorkflowContextMenu$2.actionPerformed(Unknown Source)
         at javax.swing.AbstractButton.fireActionPerformed(AbstractButton.java:2018)
         at javax.swing.AbstractButton$Handler.actionPerformed(AbstractButton.java:2341)
         at javax.swing.DefaultButtonModel.fireActionPerformed(DefaultButtonModel.java:402)
         at javax.swing.DefaultButtonModel.setPressed(DefaultButtonModel.java:259)
         at javax.swing.AbstractButton.doClick(AbstractButton.java:376)
         at javax.swing.plaf.basic.BasicMenuItemUI.doClick(BasicMenuItemUI.java:833)
         at javax.swing.plaf.basic.BasicMenuItemUI$Handler.mouseReleased(BasicMenuItemUI.java:877)
         at java.awt.Component.processMouseEvent(Component.java:6504)
         at javax.swing.JComponent.processMouseEvent(JComponent.java:3321)
         at java.awt.Component.processEvent(Component.java:6269)
         at java.awt.Container.processEvent(Container.java:2229)
         at java.awt.Component.dispatchEventImpl(Component.java:4860)
         at java.awt.Container.dispatchEventImpl(Container.java:2287)
         at java.awt.Component.dispatchEvent(Component.java:4686)
         at java.awt.LightweightDispatcher.retargetMouseEvent(Container.java:4832)
         at java.awt.LightweightDispatcher.processMouseEvent(Container.java:4492)
         at java.awt.LightweightDispatcher.dispatchEvent(Container.java:4422)
         at java.awt.Container.dispatchEventImpl(Container.java:2273)
         at java.awt.Window.dispatchEventImpl(Window.java:2713)
         at java.awt.Component.dispatchEvent(Component.java:4686)
         at java.awt.EventQueue.dispatchEventImpl(EventQueue.java:707)
         at java.awt.EventQueue.access$000(EventQueue.java:101)
         at java.awt.EventQueue$3.run(EventQueue.java:666)
         at java.awt.EventQueue$3.run(EventQueue.java:664)
         at java.security.AccessController.doPrivileged(Native Method)
         at java.security.ProtectionDomain$1.doIntersectionPrivilege(ProtectionDomain.java:76)
         at java.security.ProtectionDomain$1.doIntersectionPrivilege(ProtectionDomain.java:87)
         at java.awt.EventQueue$4.run(EventQueue.java:680)
         at java.awt.EventQueue$4.run(EventQueue.java:678)
         at java.security.AccessController.doPrivileged(Native Method)
         at java.security.ProtectionDomain$1.doIntersectionPrivilege(ProtectionDomain.java:76)
         at java.awt.EventQueue.dispatchEvent(EventQueue.java:677)
         at java.awt.EventDispatchThread.pumpOneEventForFilters(EventDispatchThread.java:211)
         at java.awt.EventDispatchThread.pumpEventsForFilter(EventDispatchThread.java:128)
         at java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:117)
         at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:113)
         at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:105)
         at java.awt.EventDispatchThread.run(EventDispatchThread.java:90)
    I get this error for Decision Tree Model only. For other models, there is no problem. How can I fix it?
    Thank you.

    After setting system language to English, be sure to restart SQL Dev to retry the test.
    Make sure you have the latest patches installed.
    Go to menu: Help-> About. In the About dialog, select the Extensions tab and check if you have Data Miner version 11.2.0.2.04.40. If not, go to menu Help->Check for Updates, and install the SQL Developer and Data Miner patches.
    SQL Dev 3.0 is not certified with Java 7.
    Thanks,
    Marat
    Edited by: mspivak on Jan 10, 2012 12:33 PM
    Edited by: mspivak on Jan 10, 2012 12:33 PM
    Edited by: mspivak on Jan 10, 2012 12:34 PM

  • Problem in datamining :Wanting package holding classes for Decision Trees

    Hello,Friends.Please,I have downloaded the package odmapi.jar several times (several releases ) but there was a problem that I did not find the class DecisionTreesSettings(or classes concerning to Decision Trees ) in the path :oracle.dmt.odm.settings.algorithm...
    Please help with giving me release of this package that holds this class....thanks

    I only have to put null in the instance variable declaration for it
    to pop up again in my application.This almost always points to you performing some initialization from a superclass constructor. I know you say this isn't happening. However, it's by far the most likely situation. You could also have threading issues, as you don't synchronize your singleton creation.
    FYI, Java initializes class instances using the following (simplified) sequence
    1) Allocate the memory and initialize it to default values (nulls for references, 0 for numbers, false for booleans). This happens before any constructors are called.
    2) For each class in the hierarchy, starting with Object, invoke the instance initializers and appropriate constructor. If a superclass constructor changes a variable defined by a subclass (typically by calling one of the subclass's methods), that value will be overwritten by any instance initializer (or constructor-time assignment) of the subclass.
    For testing, I suggest you put logging messages in the code that assigns the variable, and at the start of the subclass constructor. You should probably log the current thread ID, and it wouldn't hurt to dump a stack trace as well.
    Edit: wow, lots of posts between the time I started writing this and the time I finished ... oh well, perhaps the OP will understand instance initialization better if he/she reads this.

  • How to create a decision tree in ABAP (in SAP or not, and how????)

    Hello,
    I have to create a decision tree which has to be called within a BAPI.
    This tree finally has to return 1 single value.
    Can you tell me what's the best solution for this ?
    Should I do this with abap, or should I call a webapplication which provides this tree for me ?
    It would be great if you could provide some example-code for me.
    Kind regards,
    Gert-Jan

    I am not sure I understood what you meant by decision tree. If you are simply talking about arriving at an answer based on one or more inputs, then it is a simple IF ELSEIF ELSE statements or CASE statements.
    Can you please elaborate/give example of what you mean by decision tree?

  • Problem in Exploring the Decision Tree Model on Lesson 4 Basic Data Mining SSAS

    Hello everyone,
    I've tried to follow all the steps mentioned in Basic Data Mining Tutorials, but I've an odd problem in the Lesson 4 - Exploring the Decision Tree Model (http://technet.microsoft.com/en-us/library/cc879269.aspx).
    It's stated that "As you view the TM_Decision_Tree model
    in the Decision Tree viewer, you can see the most important attributes at the left side of the chart. “Most important” means that these attributes have the greatest influence on the outcome. Attributes further down the tree (to the right of the chart) have
    less of an effect. In this example, age is the single most important factor in predicting bike buying. The model
    groups customers by age, and then shows the next more important attribute for each age group. For example, in the group of customers aged 34 to 40, the number of cars owned is the strongest predictor after age."
    But, I got a different result from what mentioned in the tutorial. I got the number of cars owned as the most important attribute, then followed by the attribute age.
    Do you guys know why?
    Thanks in advance

    HI,
    BEGIN
       INSERT INTO DT_CA_SETTINGS_TEST (SETTING_NAME, SETTING_VALUE) VALUES
       (dbms_data_mining.TREE_TERM_MINPCT_NODE,to_char(1));
    END;
    That is not "Mode" its "NODE".
    Now Execute.
    Cheers..

  • Regenerate decision tree for transfering to unicode system.

    Hello,
    We upgraded the system to a unicode system and we're getting runtime errors using PB infotypes (pb20/pb30) for the
    generated forums in the decision tree.
    The error thrown is unicode compatability, If I make a "false" change in a certain decision tree for a field, the form is regenerated and It works.
    The problem is , that I have over 200 fields for decision trees, and I need them all regenerated.
    Someone gave me report "RPUMKG00" as a solution (activation type 1), but It does not regenerate the forms in a unicode mode.
    does anybody have any ideas?
    thanks.

    Hello,
    We upgraded the system to a unicode system and we're getting runtime errors using PB infotypes (pb20/pb30) for the
    generated forums in the decision tree.
    The error thrown is unicode compatability, If I make a "false" change in a certain decision tree for a field, the form is regenerated and It works.
    The problem is , that I have over 200 fields for decision trees, and I need them all regenerated.
    Someone gave me report "RPUMKG00" as a solution (activation type 1), but It does not regenerate the forms in a unicode mode.
    does anybody have any ideas?
    thanks.

  • Training Decision Tree: Get RSDME362

    Hi,
    I want to train my decision tree modell and get an error: RSDME362, Vorhers. Modellf. hat nur einen Wert.Daten können nicht trainiert werden.
    If I look at the statistics of the attribute which will be predicted it has two distinct values: 1.000 and 0.000.
    Does anybody knows what could be wrong??
    Thanks,
    regards

    Hi,
    I found the problem: the algorithm couldn't process values with 0. So we changed it to 1 (=yes) and 2 (=no) instead of 0 (no).
    Regards

  • What is a Decision Tree Algorithm?

    hi there,
    what is a decision tree algorithm??

    Hi,
    Decision trees are used to learn from historic data and to make predictions about the future. Prediction involves establishing rules using historic data and applying these rules to new data. These rules are displayed graphically as a hierarchy.
    Ex : Your customer data typically contains attributes such as gender, age, income, region, and occupation as well as information about whether a customer is a satisfied customer or not (possibly drawn from a survey). You can use such historic data to train a decision tree. You find out as a result that customers exhibiting certain attributes are generally satisfied customers while customers exhibiting other attributes tend to be dissatisfied customers. You can use rules determined in this way to assess the satisfaction of other customers in cases where this information is not available.
    Integration :
    The data that you use to train the model can be taken from any other system, provided that the system can extract data into SAP BW. Similarly, you can apply the identified rules to any data that has been extracted into SAP BW. In SAP BW, you can use queries to access data with known statements and then use this data to find out statements about other data.
    Regards
    Ram.

  • In cloud for service, is there a way to generate RMA's and decision trees?

    Also, where can I find more information on the field service capabilites?

    Yes it is supported via HCI configuration. C4Service provides a standard iFlow that creates a follow-up SD Order (Billing Request) from a C4Service Work Ticket. You can easily reconfigure the iFlow using HCI tools to create a SD Return Order instead of a SD Billing Request. This would then allow you to easily configure the C4Service work ticket to be a RMA capture form so that an end user can capture the products/parts as items for the SD Return Order (RMA). The end user can then release the Work Ticket to ERP which would create the RMA and ERP would take over the RMA process.
    As for the decision tree, this is supported using dynamic checklists and surveys for tickets. You can configure a dynamic survey/checklist that can have a series of questions that based on questions answered in the previous step can have different follow-up questions. These surveys/checklists can instruct the end user through a series of questions to help them diagnose a problem to ultimately instruct the end user to release the work ticket to SD for a Return Order (RMA).

  • How to find optimal multi splits for numerical attributes in Decision Tree

    Hi experts,
    I would like to know how to find optimal multi splits for numerical attributes
    in Decision Tree Learning.
    For splitting categorical attributes, gini and gain index are used and the splits that
    produce the best gain are selected.
    However, in case of numerical attributes
    how do you find the best splits?????
    Please help me !!

    Yordan's suggestion is correct, though the documentation is a bit outdated.
    You can use the single parameter + sweeper mode or define your own custom ranges in the BDT module with the Parameter Range setting
    Regards,
    aK

  • CSTVI decision tree

    Hi All,
                can any1 plz help me how to create CSTVI decision tree.
       Is it compulsory to create CSTVI, CRDVI,COVGR,EECGR decision trees&  why do we create them?
    Thanks,
    satya

    1. You can read more about the algorithm at
    http://technet.microsoft.com/en-us/library/ms175312.aspx ,
    http://technet.microsoft.com/en-us/library/cc645868.aspx
    2. You don't need to write code to use algorithms from SQL Server. If you are just getting started, I recommend you to use Excel Data Mining Add-ins to create the models
    http://www.microsoft.com/en-us/download/details.aspx?id=35578
    If you don't have Excel, you can create models from SQL Server Development Studio, but it is more difficult to use.
    What are you trying to do?
    Tatyana Yakushev [PredixionSoftware.com]
    Download Predixion Insight 3.0 - World class predictive platform for big data

Maybe you are looking for

  • Desktop Sync No Longer Works

    Thank you for reading and for any input you are willing to offer. I have Windows 7 Professional. My Outlook profile is saved on an Exchange server and my Blackberry had previously beeng syncing with Outlook in this setup for nearly a year without is

  • Digital Publishing Suite Help | What's new in this release

    This question was posted in response to the following article: http://helpx.adobe.com/digital-publishing-suite/help/whats-new-release.html

  • Problem during transport

    Dear All, Now i meet a problem during transport my ABAP in dev to prod server. Not sure is BASIS or my problem. The error is "Request after end mark" One more question is for the transaction code STMS, the status for production server should be LOCK

  • Different columns and arrangements in different Folders

    I arrange my mail into various folders. For various reasons, I need to have different columns visible and different arrangements of columns in different folders (Kind of like needing From: in InBox and To: in sent items.) I can't figure out how to se

  • How do I make sure that CS5.5 Merge clips encodes Audio at correct volume?

    Windows 7 x 64, 2 TB, 12 GB RAM, I am using the Premiere Pro CS5.5 Trial (full version), particularly because of the new Merge clips feature. The Merge clips works intuitively and easily; this is a fantastic advantage. I am creating these clips to us