NeuNet Pro Program Screens
Table Of Contents

Screens

Configure Project

Advanced Config

Data Split

SFAM Train

SFAM Browse

Confusion Matirx

BackProp Train

BackProp Browse

Scatter Graph

Time Series Graph

BackProp Scatter Graph

Normalized RMS Error Actual RMS Error Unexplained Variance Correlation Coefficient Scatter Graph
This screen provides both a numeric and graphical report showing the accuracy of your BackProp predictions. Double click anywhere on the graph to browse data from that area.

Normalized RMS Error:

  • This error indicates the Root Mean Square (RMS) error for the entire testing set.
  • This error is also called "Standard Error of Estimate".
  • The error is calculated as SquareRoot{SumOfAll[(Actual-Predicted)2] / NumberOfPredictions}
  • This calculation is performed using normalized values, so it may be stated as percent.
  • Approximately 2/3 of your predictions should lie within ± this percent of the actual target value.

Actual RMS Error:

  • This error indicates the Root Mean Square (RMS) error for the entire testing set.
  • This error is also called "Standard Error of Estimate".
  • The error is calculated as SquareRoot{SumOfAll[(Actual-Predicted) 2] / NumberOfPredictions}
  • This calculation is performed using denormalized values, so it is stated in the same units as the prediction field.
  • Approximately 2/3 of your predictions should lie within ± these units from the actual target value.

Unexplained Variance:

  • This number indicates what portion of the target value is not explained by the prediction value.
  • Unexplained Variance = Actual_Rms_Error 2 / Variance_Of_Target_Column

Correlation Coefficient:

  • This is a number between zero and one which indicates how well the prediction is correlated to the actual.
  • A value of one indicates perfect predictions.
  • A value of zero indicates no relationship between prediction and target.
  • Correlation Coefficient = SquareRoot(1 - Unexplained Variance)

Scatter Graph:

  • This is a visual representation showing the accuracy of all predictions.
  • Double click anywhere on the graph to browse data from that area.
  • The closer the scattering is to the blue diagonal line, the more accurate are the predictions.
  • The band shown by faint yellow lines indicates a certainty band, as defined by the RMS Error. Approximately 2/3 of your testing set should lie within this certainty band.
  • The sample scatter graph shown above is showing that we have greater accuracy on predicting the age of younger abalone than on older abalone.



A Complete Neural Network Development System

CorMac Technologies Inc.
34 North Cumberland Street ~ Thunder Bay ON P7A 4L3 ~ Canada
E m a i l