Loading ...
Sorry, an error occurred while loading the content.
 

NEWBY question - how to view prediction values in EasyNN

Expand Messages
  • kellymcgillivray01
    I am brand new to both Neural Networks and the EasyNN software. I have a datasest with one output and 17 numerical inputs. There are about 5600 training
    Message 1 of 3 , Feb 28, 2012
      I am brand new to both Neural Networks and the EasyNN software. I have a datasest with one output and 17 numerical inputs. There are about 5600 training records and 1300 validating records. Once the network is trained, I can see and/or export all the various measures of error, sensitivity, weights, etc. and I can see a graph of the training and validating examples.

      - Can I export a file of the actual predictions for the validation records?

      - How do I then use the trained ANN to predict other records for which I have the 17 data inputs but no actual outputs?

      - Is there a way to determine the final "equation" or set of equations that would make up the final predictions, or must I continue to use the neural network software to make predictions?

      If I am on the wrong track or my questions don't make sense, please feel free to set me straight. Thanks for any help as soon as possible!
    • Stephen Wolstenholme
      ... You can export the grid to a *.txt or *.csv file and then find the predictions for the validation rows. ... Use Query or Query from file. ... You can
      Message 2 of 3 , Feb 29, 2012
        On Tue, 28 Feb 2012 20:28:29 -0000, you wrote:

        >I am brand new to both Neural Networks and the EasyNN software. I have a datasest with one output and 17 numerical inputs. There are about 5600 training records and 1300 validating records. Once the network is trained, I can see and/or export all the various measures of error, sensitivity, weights, etc. and I can see a graph of the training and validating examples.
        >
        >- Can I export a file of the actual predictions for the validation records?

        You can export the grid to a *.txt or *.csv file and then find the
        predictions for the validation rows.

        >
        >- How do I then use the trained ANN to predict other records for which I have the 17 data inputs but no actual outputs?

        Use Query or Query from file.

        >
        >- Is there a way to determine the final "equation" or set of equations that would make up the final predictions, or must I continue to use the neural network software to make predictions?
        >

        You can export Activation value and formulae for the whole network.

        >If I am on the wrong track or my questions don't make sense, please feel free to set me straight. Thanks for any help as soon as possible!

        I will help when I can.

        Steve

        --
        Neural Network Software. http://www.npsl1.com
        EasyNN-plus. Neural Networks plus. http://www.easynn.com
        SwingNN. Forecast with Neural Networks. http://www.swingnn.com
        JustNN. Just Neural Networks. http://www.justnn.com
      • P.Dickens
        - Is there a way to determine the final equation or set of equations that would make up the final predictions, You must continue to use your trained network
        Message 3 of 3 , Feb 29, 2012
          - Is there a way to determine the final "equation" or set of equations
          that would make up the final predictions,

          You must continue to use your trained network to act as your predictor
          for any new data.

          The Equation as you put it, is Not a mathematical one. It is based on
          the 'Converged' Network that you have produced, i.e. the Trained
          Network. NN do not work in the manner, that I believe your question is
          directed. There is no end Mathematical, Algebraic nor Statistical
          formula, that is produced. If you are a Statastician or Mathematician
          and that is what you seek, then contact me again.

          You should periodically update your trained network by adding your new
          data to your existing training data. Then rerun your network to train
          again. A word of caution do not loose your original trained network,
          keep it under a seperate name. When you run a newly converged net. to
          predict with fresh data, do the same with you old or older networks and
          compare their relative accuracy with your newest one. NNets can be
          perverse & you may well find an older network will give you far better
          results than a more up to date newer one. If you are a methematician
          you probably won't like that. But that's NN's., and the wonder of the
          Dark Arts.

          P. Dickens


          ========================================
          Message Received: Feb 29 2012, 09:56 AM
          From: "kellymcgillivray01"
          To: easynn@yahoogroups.com
          Cc:
          Subject: NEWBY question - how to view prediction values in EasyNN

          I am brand new to both Neural Networks and the EasyNN software. I have a
          datasest with one output and 17 numerical inputs. There are about 5600
          training records and 1300 validating records. Once the network is
          trained, I can see and/or export all the various measures of error,
          sensitivity, weights, etc. and I can see a graph of the training and
          validating examples.

          - Can I export a file of the actual predictions for the validation records?

          - How do I then use the trained ANN to predict other records for which I
          have the 17 data inputs but no actual outputs?

          - Is there a way to determine the final "equation" or set of equations
          that would make up the final predictions, or must I continue to use the
          neural network software to make predictions?

          If I am on the wrong track or my questions don't make sense, please feel
          free to set me straight. Thanks for any help as soon as possible!


          Reply to sender
          <mailto:mcgillivray@...?subject=Re%3A%20NEWBY%20question%20-%20how%20to%20view%20prediction%20values%20in%20EasyNN>
          | Reply to group
          <mailto:easynn@yahoogroups.com?subject=Re%3A%20NEWBY%20question%20-%20how%20to%20view%20prediction%20values%20in%20EasyNN>
          | Reply via web post
          <http://groups.yahoo.com/group/easynn/post;_ylc=X3oDMTJxZjRvM2R2BF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRtc2dJZAMxNjc4BHNlYwNmdHIEc2xrA3JwbHkEc3RpbWUDMTMzMDUwOTM4MQ--?act=reply&messageNum=1678>
          | Start a New Topic
          <http://groups.yahoo.com/group/easynn/post;_ylc=X3oDMTJmdWJxc2E4BF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRzZWMDZnRyBHNsawNudHBjBHN0aW1lAzEzMzA1MDkzODE->

          Messages in this topic
          <http://groups.yahoo.com/group/easynn/message/1678;_ylc=X3oDMTM1Y3ZxMjNiBF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRtc2dJZAMxNjc4BHNlYwNmdHIEc2xrA3Z0cGMEc3RpbWUDMTMzMDUwOTM4MQR0cGNJZAMxNjc4>
          (1)
          Recent Activity:

          * New Members
          <http://groups.yahoo.com/group/easynn/members;_ylc=X3oDMTJnaDlsOTVvBF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRzZWMDdnRsBHNsawN2bWJycwRzdGltZQMxMzMwNTA5Mzgx?o=6>
          1

          Visit Your Group
          <http://groups.yahoo.com/group/easynn;_ylc=X3oDMTJmb3Vpc3E2BF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRzZWMDdnRsBHNsawN2Z2hwBHN0aW1lAzEzMzA1MDkzODE->

          <http://groups.yahoo.com/;_ylc=X3oDMTJlZTlqNGVpBF9TAzk3MzU5NzE0BGdycElkAzEwNjUzMDM3BGdycHNwSWQDMTcwNTk0ODkyMwRzZWMDZnRyBHNsawNnZnAEc3RpbWUDMTMzMDUwOTM4MQ-->

          Switch to: Text-Only
          <mailto:easynn-traditional@yahoogroups.com?subject=Change%20Delivery%20Format:%20Traditional>,
          Daily Digest
          <mailto:easynn-digest@yahoogroups.com?subject=Email%20Delivery:%20Digest>
          . Unsubscribe
          <mailto:easynn-unsubscribe@yahoogroups.com?subject=Unsubscribe> . Terms
          of Use <http://docs.yahoo.com/info/terms/>
          .


          [Non-text portions of this message have been removed]
        Your message has been successfully submitted and would be delivered to recipients shortly.