SANN - New Analysis/Deployment Startup Panel and Quick Tab

Ribbon bar. Select the Statistics tab. In the Advanced/Multivariate group, click Neural Nets to display the SANN - New Analysis/Deployment Startup Panel. Or, select the Data Mining tab. In the Learning group, click Neural Networks to display the Startup Panel.

Classic menus. On the Statistics menu, select Automated Neural Networks to display the SANN - New Analysis/Deployment Startup Panel. Or, on the Data Mining menu, select Automated Neural Networks to display the Startup Panel

The Startup Panel contains one tab: New analysis/Deployment.

Deployment.

Deploy models from previous analyses. Select this option button to load and deploy existing neural network models created from previous analyses. All files must be in PMML format and must have been created by STATISTICA Automated Neural Networks (SANN). You can select multiple models if they were created with the same analysis type (i.e., regression, classification, cluster, etc.). All neural network models saved in the PMML files must have the same number of input-target variables with matching names and order. All categorical variables must have the same number of categories and be presented in the same order.

Load network files. Click this button to display the Open dialog box, which is used to locate previously saved network files. Networks must be stored in PMML format. Options to save networks are available in the SANN - Results, SANN - Custom Neural Network, SANN - Automated Network Search (ANS), and SANN - Subsampling dialog boxes after training is completed. When you open network files using this option, the PMML code will be immediately parsed by STATISTICA Automated Neural Networks. Any failure to read a file will result in a warning/error message depending on whether at least one model was successfully loaded.

Model list data grid. This area, which is located under the Load network files button, displays a list of networks that has been successfully loaded and parsed by STATISTICA Automated Neural Networks. This list includes a sequential network ID, model names, and types of activation functions for both the hidden and output units.

PMML file list. Click this button to print a Current PMML spreadsheet, which includes the list of all the PMML network files you last tried to load, including the ones that failed to successfully parse. By inspecting this spreadsheet you can determine which files failed to load.

New analysis.

New analysis. Select this option button to start a new analysis, i.e., create new neural network models from the active data set.

Select the type of new analysis you want to perform: Regression, Classification, Time series (regression), Time series (classification) or Cluster analysis. These selections will constrain the types of variables you can select as input and target (independent and dependent) variables, to conform to the respective type of analysis.

Regression. Regression analysis is usually concerned with predicting one or more continuous variables with a set of inputs. Select this analysis type when your target (dependent) variables of interest are continuous in nature (e.g., weight, temperature, height, length, etc.).

Classification. Classification analysis is concerned with finding a class membership for a set of input variables. The class membership is determined by the categorical levels of the target variable. Select this analysis type when your target (dependent) variable is categorical in nature (e.g., gender). Note that for classification analysis, you can only specify one target.

Time series (regression). Select this analysis type when your target (dependent) variables are continuous in nature, and may involve lagged (over time) predictions. When this problem type is selected, the variable selection dialog will permit selection of target variables without requiring additional input variables for the network. You may, however, select only one target variable when no inputs are given.

Time series (classification). Select this analysis type when your target (dependent) variable is categorical in nature, and may involve lagged (over time) predictions. When this problem type is selected, the variable selection dialog will permit selection of target variables without requiring additional input variables for the network. Note that for time series (classification) analysis, you can only specify one target.

Cluster analysis. Select this analysis type to perform unsupervised learning based on the Kohonen algorithm, to detect clusters in the data.

OK. Click OK to begin the chosen analysis, as specified on the Startup Panel.

Cancel. Click the Cancel button to close the SANN - Analysis/Deployment Startup Panel without taking any action.

Options. See Options Menu for descriptions of the commands on this menu.

Open Data. Click the Open Data button to display the Select Data Source dialog box, which is used to choose the spreadsheet on which to perform the analysis. The Select Data Source dialog box contains a list of the spreadsheets that are currently active.