Noob Question: Flowing Input Dataset Row/Record IDs Through to the Score Model Output

I've got a noob question that I'm having difficulty finding the answer to:  if my input dataset contains row/record IDs and I want to flow these through to the Score Model module output but *don't* want to include this column in model training (because
it's just an arbitrary row tag), how do I go about accomplishing this?  
I am able to Project Columns and eliminate the IDs entirely early in the process but then can't really determine which input row a prediction is associated with because I've lost the IDs.  I'm not quite sure how to accomplish what I'm trying to do via
the Metadata Editor, either.
Surely somebody else has dealt with this same question.  If the solution is implemented somewhere in a Gallery experiment that you'd like to point me at, too, that would also be just as good as a written explanation.
Thanks in advance,
Kevin

Hey Kevin!
Sorry for making this process unclear :( Azure ML has a notion of Feature Columns
(by default everything is a Feature), that is columns/variables that are intended to be treated as part of the observation vector. You can use the Metadata Editor
and (after selecting the column(s) you want), mark the column as
Fields -> Clear Feature. This will have the semantics you desire - data attached to the input data while not being used for training.
Regards,
AK

Similar Messages

  • Is the task-flow input parameter class value being ignored by the framework

    hi
    Although I don't remember where, I think I read or heard about the task-flow input parameter class value that it is "being ignored by the framework".
    One example of this could be what is currently in the UI Shell sample application available
    at http://www.oracle.com/technetwork/developer-tools/adf/uishellapp-134633.zip
    It has in its task-flow in "flows\second.xml" an input parameter configured like
      <task-flow-definition id="second">
        <!-- ... -->
        <input-parameter-definition id="__3">
          <name id="__2">tabContext</name>
          <value>#{pageFlowScope.tabContext}</value>
          <class>oracle.ui.pattern.dynamicShell.TabContex</class>
          <required/>
        </input-parameter-definition>
        <!-- ... -->
      </task-flow-definition>Notice the missing "t" at the end of the class name "oracle.ui.pattern.dynamicShell.TabContex".
    Still this task-flow configuration does allow to use expressions like "#{pageFlowScope.tabContext.selectedTabIndex}", which suggests that the class value is indeed "being ignored by the framework".
    (see also forum thread "does the UI Shell sample break encapsulation in its task-flows"
    at does the UI Shell sample break encapsulation in its task-flows )
    All relevant references (bug numbers, documentation, blogs) explaining about this task-flow input parameter class value are welcome.
    If no such references exist, maybe someone from Oracle can give some feedback.
    many thanks
    Jan Vervecken

    fyi
    The feedback below was posted in the forum message
    at Re: does the UI Shell sample break encapsulation in its task-flows
    Richard Wright wrote:
    That appears to be a bug for which I am seeking confirmation.There is an enhancement request (i.e., 9377487) submitted over a year ago, that describes this behavior. It is listed as an ER as our documentation (e.g., Fusion Developer's Guide for Oracle ADF, online help ) does not specifically state that Java class for the input parameter definition is checked for type. Further anything using EL is untyped. So the expectation that types are enforced is somewhat puzzling to the development team.
    However, there is an acknowledgment within circles of the development organization, that some might come to have this expectation. One trigger might be the definition itself. The other is the existence of a design time audit on the definition. For example, if the class is specified inaccurately (e.g., typo), there is an audit warning to report "not found."
    Under consideration are a number of framework proposals to check for type without breaking backward compatibility or adding overhead to the production environment.
    In the interim, the following recommendations are given.
    Unless this breaks or can break your app in some way, no need exist for change. Nevertheless, if this is a real concern, write a test or consider a java assertion. By default, assertions are disabled at runtime. There is no need to override that default.I have been able to find enhancement request 9377487, "TASK FLOW PARAMETER VALUES NOT VALIDATED ACCORDING TO CLASS", on My Oracle Support. It currently has "Status 15 - To Internal (Oracle) Review" and "Updated 16-Dec-2010".
    In the context of service request 3-4185839067 recently bug 12838099, "THE TASK-FLOW INPUT PARAMETER CLASS VALUE BEING IGNORED BY THE FRAMEWORK", has been filed and it currently still has "Status 16 - Bug Screening/Triage".
    - about "... the expectation that types are enforced is somewhat puzzling to the development team ..."
    Hmm, strange ... so what are people supposed to expect when there is an option to configure a class name for a task-flow input parameter?
    regards
    Jan

  • Map input schema with highest id value in the node to output schema

    <job job_reference="4444">
    <jobnotes/>
    <appointments>
      <appointment id="1" slot_id="1" slot_ref="0830-1030" slot_start_time="08:30" slot_end_time="10:30" slot_description="0830-1030" date="19/02/2014" time="" tk_rowid="7367"
    kept="" reason_code="" narrative="" status="DEL" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:09" amended_by="" amended_date="19/02/2014 11:58:15"
    apb_exported="N" />
      <appointment id="2" slot_id="2" slot_ref="1030-1230" slot_start_time="10:30" slot_end_time="12:30" slot_description="1030-1230" date="19/02/2014" time="" tk_rowid="7368"
    kept="" reason_code="" narrative="" status="ENT" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:38" amended_by="" amended_date="19/02/2014 11:58:38"
    apb_exported="N" />
    </appointments>
    </job>
    The above is the input schema. I need to map the elements in appointment node with highest ID to the elements in the output schema. Of the two appointment nodes above, I should only pick up one node with highest ID and map the elements of it to
    the output schema.

    My apologies for not giving the input schema right.
    <jobs>
    <job job_reference="4444">
    <jobnotes/>
    <appointments>
      <appointment id="1" slot_id="1" slot_ref="0830-1030" slot_start_time="08:30" slot_end_time="10:30" slot_description="0830-1030" date="19/02/2014" time="" tk_rowid="7367"
    kept="" reason_code="" narrative="" status="DEL" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:09" amended_by="" amended_date="19/02/2014 11:58:15"
    apb_exported="N" />
      <appointment id="2" slot_id="2" slot_ref="1030-1230" slot_start_time="10:30" slot_end_time="12:30" slot_description="1030-1230" date="19/02/2014" time="" tk_rowid="7368"
    kept="" reason_code="" narrative="" status="ENT" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:38" amended_by="" amended_date="19/02/2014 11:58:38"
    apb_exported="N" />
    </appointments>
    </job>
    <job>
    <job job_reference="1111">
    <jobnotes/>
    <appointments>
      <appointment id="1" slot_id="1" slot_ref="0830-1030" slot_start_time="08:30" slot_end_time="10:30" slot_description="0830-1030" date="19/02/2014" time="" tk_rowid="7367"
    kept="" reason_code="" narrative="" status="DEL" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:09" amended_by="" amended_date="19/02/2014 11:58:15"
    apb_exported="N" />
      <appointment id="2" slot_id="2" slot_ref="1030-1230" slot_start_time="10:30" slot_end_time="12:30" slot_description="1030-1230" date="19/02/2014" time="" tk_rowid="7368"
    kept="" reason_code="" narrative="" status="ENT" scheme_id="1" day="4" created_by="" creation_date="19/02/2014 11:58:38" amended_by="" amended_date="19/02/2014 11:58:38"
    apb_exported="N" />
    </appointments></job></jobs>
    I should pick up details of the appointment with Higesht ID for every job. For every job one record should exist as the ouput.
    With the existing mapping, i am getting just one record as output no matter how many input jobs are there.
    Input schema has got multiple jobs and within each job there are many appointments. So in each job I should pick up the one
    with highest ID

  • Noob question, can't rotate object clockwise. Please see the pictures below.

    Hello everybody!
    I'm stuck at rotating objects as I tried to follow a tutorial.
    I think the answer is very simple, but because of my weak English, i can't even figure it out how to search for it on Google. tio
    So my question is how to rotate objects like this:
    Instead of this (That's how i can only rotate this object.):
    Thanks in advance!

    zollre,
    - before the value gives you clockwise. Basically, rotate angles are counterclockwise (with an implied + before the value).
    -20 degrees corresponds to 340 degrees (almost a full round counterclockwise).

  • Include records IDs in prediction

    Hi everybody, I´m doing some binary prediction experiments.  I haven't included IDs in my training datasets because I think that if I use them the algoritm will use that field in the learning process, which is not good as I know.
    But as scoring datasets I use data with the same columns, because as I have been told the Score model ask you to use the same columns that were used by the Training Model. The problem is that the resulting scored dataset doesn't have IDs, so I can´t identify
    each predicted record because they don´t have the original IDs.
    My question is: 
    Can I use some columns in the learning model and use those columns in the Score model but adding some extra columns, like IDs ?
    If that is not possible is there another option ? 

    Hi!
    Yes! The Score Module will simply ignore any extra columns that the model was not trained on.
    But there's a "better" way in that you can use the Metadata Editor
    and set the Fields -> Clear Feature. This will unmark a column as an observation vector element and will pass through Train unused!
    Hope this helps!
    Regards,
    AK

  • JFM recording and streaming at the same time

    Hello Everyone,
    I was hoping someone could help me with using a webcam device for sending and recording at the same time.
    At this moment I'm only able to record. I tried to make another processor, and put the input of the streaming processor
    to the one that is recording, but problem is the streaming processor outputs not a streaming format. If I could stream
    with RTP I could embed a player in a website and view from there, while recording at the same time.
    Could somebody explain to me what do to here, or provide me a snippet of code, I really need this to work.
    I created this code, but this only records.
    import java.io.IOException;
    import javax.media.*;
    import javax.media.format.*;
    import javax.media.protocol.*;
    public class Record {
         public Processor createProcessor(){
              Format formats[] = new Format[2];
              formats[0] = new AudioFormat(AudioFormat.IMA4);//IMA4
              formats[1] = new VideoFormat(VideoFormat.CINEPAK);//CINEPAK
              FileTypeDescriptor outputType = new FileTypeDescriptor(FileTypeDescriptor.QUICKTIME);
              Processor p = null;
              try {
                   p = Manager.createRealizedProcessor(new ProcessorModel(formats,outputType));
                   } catch (IOException e) {
                        System.out.println("Error! 1");
                   System.exit(-1);
                   } catch (NoProcessorException e) {
                        System.out.println("Error! 2");
                   System.exit(-1);
                   } catch (CannotRealizeException e) {
                        System.out.println("Error! 3");
                   System.exit(-1);
              return p;
         public static void main(String [] args) {
              Record record = new Record();
              Processor p = record.createProcessor();
              DataSink sink;
              MediaLocator dest = new MediaLocator("file://newfile.mpeg");
              try{
                   sink = Manager.createDataSink(p.getDataOutput(), dest);
                   sink.open();
                   p.start();
                   sink.start();
                   try {
                   Thread.currentThread().sleep(6000);
                   } catch (InterruptedException ie) {
                   p.stop();
                   p.close();
                   Thread.currentThread().sleep(6000);
                   sink.close();
              } catch (Exception exception) {
                   System.out.println("error!");
              System.out.println("finished");
    Kind regards,
    Bluesboy89

    Take a look at the following example. It demonstrates how to do 2 things with the same input (namely view and record)
    [http://java.sun.com/javase/technologies/desktop/media/jmf/2.1.1/solutions/JVidCap.html]

  • NOOB Question: Insert a line break ( br or p )

    Hi,
    I try my first steps with webdynpro and find it a bit crazy.
    I now have created my first webapp and it rules but I have 2 questions:
    1. I have 10 elements, which are displayed all abreast, but I want at some elements a line break (<br> or <p>, I only have the solution with FormattedTextView where I can insert html)
    2. How I can add JavaScript-Functions and otherwise <html>-code? CSS-Tags and so on... Have I actually influence of the generated HTML?
    sorry for this noob-questions.
    Christian

    Hi
    In Web Dynpro You can not write HTML , Java Script or insert any CSS code.
    It is not possible.
    If you want a line break between UI elements, you should follow Layouts
    you have different types of Layouts
    1. Matrix Layout
    2. Grid Layout
    3. Row Layout
    4. Flow Layout
    You can control your layout only using the above options.
    Abhi

  • Error in entering a new characteristic value using input ready row in WAD

    Hi Guys,
    I have created input ready query with input ready row option to enter both characteristic and key figure values.  I have created info object without master data tick in Master data/text tab  to enter free text which inturn will create a new record in master data. It is working in Bex analyzer, here I can enter any value to create a  new entry in the master data but where as in WAD  I am getting error message reading invalid values. In WAD I am restricted to choose only available master data entries. Do I miss any settings in WAD. Your help will be highly appreciated.

    Hi,
    the feature you are requesting is called "Master Data Planning". Unfortunately it is not supported by SAP. Please use the term "master data planning " to search this forum and you will find some threads about this issue.
    Hope this helps
    Matthias Nutt
    SAP Consulting Switzerland

  • How to break one row records into 3 row records

    I have two tables which consists of date records,
    table1
    first_date second_date
    01-feb-10 01-mar-10
    tabl2
    new_date old_date
    01-jan-10 01-aug-10
    if first table dates falls between new_date and old_dates, i need to showit as three row records instead of1 row in table2

    Hi,
    1001383 wrote:
    I have two tables which consists of date records,
    table1
    first_date second_date
    01-feb-10 01-mar-10
    tabl2
    new_date old_date
    01-jan-10 01-aug-10
    if first table dates falls between new_date and old_dates, i need to showit as three row records instead of1 row in table2Do both first_date and second_date have to fall between old_date and new_date, or is it enough if either one does?
    What do you want to see on those 3 rows?
    Depending on your data and your requirements, here's one way:
    WITH     cntr     AS
         SELECT     LEVEL     AS n
         FROM     dual
         CONNECT BY     LEVEL     <= 3
    SELECT  t2.*
    FROM          table2     t2
    CROSS JOIN     cntr
    WHERE   EXISTS (
    '              SELECT  1
                  FROM    table1     t1
                  WHERE   t1.first_date     BETWEEN t2.old_date AND t2.new_date
                  OR         t1.second_date     BETWEEN t2.old_date AND t2.new_date
    ;I hope this answers your question.
    If not, post a little sample data (CREATE TABLE and INSERT statements) for both tables, and the results you want from that data.
    Include examples of any special situations you need to handle, such as the same row in table2 matching several rows in table1.
    See the forum FAQ {message:id=9360002}

  • Input-Ready row with characteristic in column

    Hi,
    I have several workbooks based on queries with the property of having an Input-Ready row (context of integrated planning).
    These always worked fine, but after a bex upgrade only some of the queries continued to display the input-ready row at the bottom.
    A reason why only sometimes it works, has been asked to SAP.
    SAP refers to:
    http://help.sap.com/saphelp_nw70/helpdata/en/43/f234619e3c4c5de10000000a155369/frameset.htm
    which says:
    You cannot use characteristics in the columns and can use one structure only.
    My queries mostly contain 0fiscper and 0amount in the columns.
    Following the above logic, the input-ready row never should appear since I display the amount over 12 months (=characteristic in column).
    BEx:
    Bi AddOn 7.X (based on 7.10)
    Support Package 8, Patch 1
    additional info:
    - All queries were designed in 7.x.
    - Same result in BEx Analyzer without using Workbook.
    Question:
    - What could be the reason that in some cases the input-ready row still appears, and in others not?
    - Any suggestions for workaround?
    Thx in advance!
    Rgds, Sam

    I beleive if you post it in "Business planing" queue you will get good response...
    Business Planning

  • Since Yosemite, Snoball mic not recognized; regular input doesn't record either

    I desperately need to get a video to a small group of students in India who are doing work for me.
    I'm using Screenflow, which I know Apple doesn't support. But even in Quicktime and Garageband, I can't get the computer to recognize my Snoball mic. I can't find any way to record for these kids, because the sound is so nonexistent. I can't even record with regular input.
    I think my computer may be 2.5 to 3 years old. Is it too old for Yosemite? Is there any way I can roll it back to whatever came before? Or is the computer ruined? Recording is one of the main things I need to do daily. And I've been non-funcrioning since before Christmas.
    It was fine prior to upgrading. I don't know why I did. I should have known better. I really, really should have. I'm so mad at myself I could --- ahhh, ****, what the ****!
    I've read tons of discussions on sound failures with Yosemite, but none matched my issue. Anybody have suggestions? Already reset the PRAM.

    OK, after another couple of hours of searching I found this:
    https://discussions.apple.com/thread/4219793
    This solved my question by stooooooober  on Aug 20, 2012 1:13 PM
    OK! Thanks everyone for all of the help, but I just got in touch with Blue support, and it seems that the way they program their devices, the computer is only able to recognize one at a time. You can send it in and have them reprogram it so your computer can differentiate them.
    Weird work around, but I guess that at least answers the question of why its not working.
    I have a support request in to Blue, but I'm going to be upset if I have to buy another, different type, mic to get multi track recording. If that's the case, it will definitely be a different brand...

  • Previous record values displaying in the Group Footer row in the report.

    Hi Friends,
    I have 3 tables
    TableA:(PERNR BEGDA ENDDA are key fields)
    PERNR BEGDA    ENDDA      WERKS
    10001 1/1/2010 12/31/9999 1001
    TableB:(PERNR BEGDA ENDDA SUBTY are key fields)
    PERNR BEGDA    ENDDA       SUBTY  TYPES
    10001 1/1/2010 12/31/9999   01          COMS1
    10001 1/1/2010 12/31/9999   02         COMS2
    TableC:(PERNR BEGDA  ENDDA are key fields)
    PERNR BEGDA  ENDDA        AMNT
    10001 2/2/1997 4/3/2010      1000
    10001 4/4/2010 12/31/9999  2000
    I have joined these table by the key 'PERNR'( like A->B and A->C)
    Groped by the same key 'PERNR'
    My result rows in the report:
    PERNR BEGDA    ENDDA      WERKS  SUBTY  TYPES  BEGDA    ENDDA      AMNT
    10001 1/1/2010 12/31/9999 1001       01         COMS1  2/2/1997 4/3/2010    1000
    10001 1/1/2010 12/31/9999 1001       02         COMS2  2/2/1997 4/3/2010    1000
    10001 1/1/2010 12/31/9999 1001       01         COMS1  4/4/2010 12/31/9999 2000
    10001 1/1/2010 12/31/9999 1001       02         COMS2  4/4/2010 12/31/9999 2000
    But in the report format is like this in the Group Footer row:
    PERNR: 10001                     WERKS:1001
    SUBTY: 02                          AMNT :2000 (Current date)
                                                AMNT :1000 (Previous record)
    I created the format with "Previous" function
    but in the report it is giving the '2000' instead of '1000' in the AMT field
    Please help me, can i use the for loop in the report? or any sugessions?
    Thanks in advance.
    Regards,
    Venkata

    A group footer will dis[lay contents of the last record. You need to sort records to ensure that the last record contains link to your dynamic image.
    Ian

  • I want to be able to use airplay to stream audio to another iOS device, and then use that audio stream to be used when recording video instead of the built in microphone / microphone input. Is this possible?

    I want to be able to use airplay to stream audio to another iOS device, and then use that audio stream to be used when recording video instead of the built in microphone / microphone input. Is this possible?

    A third-party app probably cannot obtain a stream from another app. To the best of my knowledge, such a capability is not provided in the software development kit, apps being "sandboxed" from each other and so allowed to communicate only in very specific and limited ways.
    I'm not completely sure what you mean by "limitations on video capture". An iPhone, to the best of my knowledge, can natively record video only through it's built-in camera, and audio while doing video recording only through the built-in microphone or mic/headphone jack. There might be a video recording app that would allow audio input from an external device connected to the dock connector, but I'm not sure.
    Regards.

  • Obj or view to flatten "cell" records into "row" record

    I have ProdPrice records which are essentially "cells" in a price matrix.
    create table ProdPrice {
    ProdNum number (12,0),
    Quality char(1),
    Condition char(1),
    Price number(15,3),
    ... (other stuff)
    with primary key (ProdNum, Quality, Condition)
    where Quality is in [ 0, 1, 2, 3 ] and Condition in [ 0, 1, 2 ]
    I want to create a view or object view representing one "row"
    in the matrix with that "row" containing all prices with the same Quality. Essentially I want to "flatten" the cells into Quality "rows"
    for viewing and editing.
    An example Quality "row" type might look like:
    create type ProdPriceQualityRow {
    ProdNum number (12,0),
    Quality char(1),
    PriceCondition0 number(15,3),
    PriceCondition1 number(15,3),
    PriceCondition2 number(15,3),
    PriceCondition3 number(15,3)
    I think I now need to
    create view ProdPriceQualityRowView of type ProdPriceQualityRow
    with object identifier (ProdNum, Quality) as ....
    but I'm not sure how to complete the view / object and its associated instead of trigger.
    Does anyone have any examples of how to create views / objects / triggers that work like this?
    I've read most of the Oracle Application Developers Guide - Object Relational Features but can not seem to warp their examples into the above.
    Thanks
    R.Parr
    Temporal Arts

    For examples goto askTom.Oracle.com and search for "pivot table".

  • Bj or view to flatten "cell" records into "row" record

    I have ProdPrice records which are essentially "cells" in a price matrix.
    create table ProdPrice {
    ProdNum number (12,0),
    Quality char(1),
    Condition char(1),
    Price number(15,3),
    ... (other stuff)
    with primary key (ProdNum, Quality, Condition)
    where Quality is in [ 0, 1, 2, 3 ] and Condition in [ 0, 1, 2 ]
    I want to create a view or object view representing one "row"
    in the matrix with that "row" containing all prices with the same Quality. Essentially I want to "flatten" the cells into Quality "rows"
    for viewing and editing.
    An example Quality "row" type might look like:
    create type ProdPriceQualityRow {
    ProdNum number (12,0),
    Quality char(1),
    PriceCondition0 number(15,3),
    PriceCondition1 number(15,3),
    PriceCondition2 number(15,3),
    PriceCondition3 number(15,3)
    I think I now need to
    create view ProdPriceQualityRowView of type ProdPriceQualityRow
    with object identifier (ProdNum, Quality) as ....
    but I'm not sure how to complete the view / object and its associated instead of trigger.
    Does anyone have any examples of how to create views / objects / triggers that work like this?
    I've read most of the Oracle Application Developers Guide - Object Relational Features but can not seem to warp their examples into the above.
    Thanks
    R.Parr
    Temporal Arts

    For examples goto askTom.Oracle.com and search for "pivot table".

Maybe you are looking for