The viewer for the residual plot of a regression model has the following tabs:
This tab displays the residual plot.
A residual is the difference between the actual value and the predicted value for a model. If the predictions of a model are perfect, then the residuals are all 0. In most situations, there will be cases for which the predicted value is more accurate and cases for which the predicted value is less accurate. The residual plot lets helps you identify such cases.
The residual plot is a scatter plot where the X-axis is the predicted value and the Y-axis is the residual. You can change the assignment of axes.
The residual plot values are dots in a band above and below 0. The closer the dots are to 0, the better the predictions. If X-axis is actual values, dots with Y coordinates above 0 indicate predictions that are too high; dots with Y coordinates below 0 indicate predictions that are too low. You may find that predictions are better for certain X values than for others; for example, you might find that predictions are good for 0>X<1, high for 1>X<2, low for X>3.
The Residual Plot tab has two tabs:
This tab displays the scatter plot of residuals. You can change the following:
If the X-axis represents predicted values (the default), then a dot with Y coordinate +100 indicates that the predicted value is 100 more than the actual value, that is, the predicted value is too large.
If the X-axis represents actual values, then a dot with Y coordinate -100 indicates that the predicted value is 100 less than the actual value, that is, the predicted value is too small.
This tab displays the data for the residual plot. The apply output is displayed in three columns, Case ID, actual target value, and PREDICTION (the predicted value).
This tab describes the task used to create the plot. It shows when the task ran, what input was used, and what output was created.
Copyright © 2006, 2008, Oracle. All rights reserved.