The Analyze Model window contains two tabs: the Graph tab and the Models tab.
...
Graph Tab
The Graph tab of the Analyze Model window shows a condition versus age graph, which contains both a curve and data points:
- The curve results from the mathematical equation underlying the model. In a sense, it is the ideal curve for this particular model.
The data points are derived from regression analysis for just the selected condition attribute and node (and child nodes, if any). (This may or may not be the entire data set.) The data points can be colored black or gray, with the black data points denoting outliers.
Note Note: Outliers are only plotted if the model selected in the Models tab is the one that has the 0-Selected value in the Model Run column.
...
- Select Edit Outliers, and the User-Defined Outliers window opens.
- Right-click in the window and choose Insert. The outlier criteria are additive. The example below shows a portion of this window where young, poor data points (those with an index value less than 50 or those less than five years old) and old, good data points (those with an index value greater than 85 or those older than 20) are defined as outliers:
- Once the outlier definitions are as desired, click OK to close the User-Defined Outliers window.
- You can then include the new outlier definitions in the model selected for detailed analysis by clicking the Models tab. In the Models tab, the system will display a message to advise that the outlier definitions have changed.
- Click Yes to continue, and then the new outlier definitions will be saved and used in displaying data points on the Graph tab.
...
...
Models Tab
The Models tab of the Analyze Models window is where you perform all regression analysis (including the selection of what model form to use in the regression analysis) and where you identify which models to graph in the Graph tab.
The tab provides two panes (Current Models on the left and Model Development on the right) along with the Model Type field. This field identifies the model form whose models will be shown in the Current Models pane. The model type is selected from a drop-down list, which currently has seven types. See Model Equations and Regression Methods for more information on the different model types.
...
- Calculate Excluding Outliers: This command is only available for the model with a Model Run type of 0-Selected provided that data points exist in the data set for the model. When available, it runs a regression analysis for the model's data set with the identified outliers excluded. (Outliers are identified by either using the Define Outliers command in this pane or by using the Edit Outliers command in the Graph tab.)
- Calculate Regression Models: This command is only available for the model with a Model Run type of 0-Selected provided that data points exist in the data set for the model. When available, it performs linear regression analyses for the linear, exponential, and hyperbolic model types using the entire data set. It then creates three new records in this pane to contain the results of the analyses, with each record being assigned a Model Run type of 1-Regression.
- Define Outliers: This command is only available for the model with a Model Run type of 0-Selected provided that data points exist in the data set for the model. When available, it identifies outliers based upon the currently set Standard Error (SE) Coefficient value shown in the SE COEF column. After identifying the outliers, it then places the outlier count in the record with a Model Run type of 0-Selected. (The identified outliers are then excluded from regressions run using the Calculate Excluding Outliers command.)
- Find Closest Models: This command is available for all Model Run types, but only if data points exist in the data set for the model. It performs a "least squares" quasi-regression analysis for each model type using the entire data set. It then inserts six records in the pane, one for each model type, to contain the results of the analysis. Each record is assigned a Model Run type of 4-Least Squares.
- Move for Analysis: This command is only available for models that are not designated with a Model Run type of 0-Selected. When available, it copies the values of the record that you right-clicked into the record designated 0-Selected. Once copied, the modified 0-Selected record can then be regressed with outliers and/or added to the set of permanently available models.
- Return Current Model: This command is only available for the model with a Model Run type of 0-Selected provided the model is not already permanently available. When available, it applies the model of the 0-Selected record to the currently selected node in the Performance Models window and also includes the model in the list of permanently available models (that is, those models shown in the Setup Models window and shown in the left pane of the Models tab).
- Set Standard Error: This command is only available for the model with a Model Run type of 0-Selected provided that data points exist in the data set for the model. When available, it sets the Standard Error (SE) Coefficient value shown in the SE COEF column.
...