Evaluate regression model performance
To download the table as a csv-files click the top download button on the right of your screen. To download the plots at a png file click the lower download icon on the right of your screen.
The numeric outcome, or response, variable of interest.
Select one or more variables that can be used to predict the value of the response variable. This could be a variable or predicted values from a model (e.g., from a regression estimated using Model > Linear regression (OLS) or a Neural Network estimated using Model > Neural Network (ANN)).
filter is active (e.g., set in the Data > View tab) generate results for
Validation data, or
Both training and validation data. If no filter is active calculations are applied to all data.
Predictions were derived from a linear regression and an neural network with two nodes in the hidden layer on the
diamonds data. The variables
carat were log-transformed prior to estimation.The data is available through the Data > Manage tab (i.e., choose
Examples from the
Load data of type drop-down and press
Load examples). The predictions shown below were generated in the Predict tab.
The test statistics show a small, but consistent, advantage for the ANN.