| Use previous input description
|
Select this check box to use the variable selection specified in the Select dependent variables and predictors node, accessed via the ribbon bar or by double-clicking the data source node in the workspace. Note that the previous
Select Cases and
Weight specifications will be used also. If case selection conditions and/or weights were previously specified, you can view those specifications from this workspace dialog box, but you cannot change them (the options will be dimmed).
|
| Variables
|
Click the Variables button to display the variable selection dialog box. You can select Continuous targets and Continuous and/or Categorical inputs. Note that you should only perform time series analysis when your data involve lagged (over time) predictions.
Note: For time series analysis, selecting input variables is optional. In that case a Statistica SANN time series analysis (whether regression or classification) will map future values of the target variable on its past measurements. In other words
SANN will predict future values of the target variable from its past.
|
| Analysis variables (present in the dataset)
|
This group box displays the variables currently selected for the analysis.
|
| Strategy for creating predictive models
|
SANN provides neural network building strategies that can be used in generating your models. The options available to you during training will depend on the selection you make here.
|
| Automated network search (ANS)
|
Select this option button to enable the Automated Network Search (ANS), which is used for creating neural networks with various settings and configurations with minimal effort. ANS helps you create and test neural networks for your data analysis and prediction problems. It designs a number of networks to solve the problem and then selects those networks that best represent the relationship between the input and target variables. ANS is not applicable when deploying models since no training of neural networks is required.
|
| Custom neural networks (CNN)
|
Select this option button to use the Custom neural networks (CNN) strategy. In contrast to ANS, the Custom Neural Networks (CNN) tool enables you to choose individual network architectures and training algorithms to exact specifications. You can use CNN to train multiple neural network models with exactly the same design specifications but with different random initialization of weights. As a result, each network will find one of the possible solutions posed by neural networks of the same architecture and configurations.
|
| Subsampling (random, bootstrap)
|
Select this option button to use the
Subsampling (random, bootstrap) strategy. This tool enables you to create an ensemble of neural networks based upon multiple subsamples of the original data set. Options for the
Subsampling (random, bootstrap) strategy are available on the
Subsampling tab.
|
| MD handling (inputs)
|
Specify how to treat cases with missing values (in the input variables of the selected models).
|
| Casewise
|
When this option button is selected, any cases with missing values are omitted when generating results. Cases with missing target values are labeled as "Missing” and used to form the "Missing” sample. The missing sample consists of data cases with one or more missing target values.
|
| Mean substitution
|
The mean substitution procedure is used to "patch" missing values before training or executing the network. When this option button is selected, missing values are replaced with the training sample mean. Note that this option is applicable only to continuous variables, i.e., the
Mean substitution option button will be disabled when there are no continuous inputs in the analysis, and that for classification tasks, the option cannot be applied to the target variable, in which case all cases with missing targets will be labeled as "missing," which means a case with missing target value. Such cases are grouped in
SANN as the missing sample and can be used for fixing the basis functions of the RBF neural networks and for making predictions.
Note: The mean substitution option will always compute the simple arithmetic mean to replace missing data, even when weights are in effect. Weights in
SANN are used ("interpreted") as measures of case "importance," i.e., they will affect the estimation of neural network parameters themselves. If the intention of weights is to compute a weighted mean (e.g., a population average computed using weights) to replace missing data in the input file, use option Data - Filter/Recode - Process Missing Data to replace missing data values with weighted means.
Options / C / W. See Common Options.
|
|
|
| OK
|
Click the
OK button to accept all the specifications made in the dialog box and to close it. The analysis results will be placed in the
Reporting Documents node after running (updating) the project.
|
