Output Options (Temporal Causal Modeling)
These options specify the content of the output. Options in the Output for targets group generate output for the targets that are associated with the bestfitting models on the Series to Display settings. Options in the Output for series group generate output for the individual series that are specified on the Series to Display settings.
 Overall model system

Displays a graphical representation of the causal relations between series in the model system. Tables of both model fit statistics and outliers for the displayed targets are included as part of the output item. When this option is selected in the Output for series group, a separate output item is created for each individual series that is specified on the Series to Display settings.
Causal relations between series have an associated significance level, where a smaller significance level indicates a more significant connection. You can choose to hide relations with a significance level that is greater than a specified value.
 Model fit statistics and outliers
 Tables of model fit statistics and outliers for the target series that are selected for display. These tables contain the same information as the tables in the Overall Model System visualization. These tables support all standard features for pivoting and editing tables.
 Model effects and model parameters
 Tables of model effects tests and model parameters for the target series that are selected for display. Model effects tests include the F statistic and associated significance value for each input included in the model.
 Impact diagram

Displays a graphical representation of the causal relations between a series of interest and other series that it affects or that affect it. Series that affect the series of interest are referred to as causes. Selecting Effects generates an impact diagram that is initialized to display effects. Selecting Causes generates an impact diagram that is initialized to display causes. Selecting Both causes and effects generates two separate impact diagrams, one that is initialized to causes and one that is initialized to effects. You can interactively toggle between causes and effects in the output item that displays the impact diagram.
You can specify the number of levels of causes or effects to display, where the first level is just the series of interest. Each additional level shows more indirect causes or effects of the series of interest. For example, the third level in the display of effects consists of the series that contain series in the second level as a direct input. Series in the third level are then indirectly affected by the series of interest since the series of interest is a direct input to the series in the second level.
 Series plot
 Plots of observed and predicted values for the target series that are selected for display. When forecasts are requested, the plot also shows the forecasted values and the confidence intervals for the forecasts.
 Residuals plot
 Plots of the model residuals for the target series that are selected for display.
 Top inputs
 Plots of each displayed target, over time, along with the top 3 inputs for the target. The top inputs are the inputs with the lowest significance value. To accommodate different scales for the inputs and target, the y axis represents the z score for each series.
 Forecast table
 Tables of forecasted values and confidence intervals of those forecasts for the target series that are selected for display.
 Outlier root cause analysis
 Determines which series are most likely to be the cause of each
outlier in a series of interest. Outlier root cause analysis is done for each
target series that is included in the list of individual series on the Series
to Display settings.
 Output

 Interactive outliers table and chart
 Table and chart of outliers and root causes of those outliers for each series of interest. The table contains a single row for each outlier. The chart is an impact diagram. Selecting a row in the table highlights the path, in the impact diagram, from the series of interest to the series that most likely causes the associated outlier.
 Pivot table of outliers
 Table of outliers and root causes of those outliers for each series of interest. This table contains the same information as the table in the interactive display. This table supports all standard features for pivoting and editing tables.
 Causal levels
 You can specify the number of levels to include in the search for the root causes. The concept of levels that is used here is the same as described for impact diagrams.
 Model fit across all models
 Histogram of model fit for all models and for selected fit statistics. The following
fit statistics are available:
 R square
 Goodnessoffit measure of a linear model, sometimes called the coefficient of determination. It is the proportion of variation in the target variable explained by the model. It ranges in value from 0 to 1. Small values indicate that the model does not fit the data well.
 Root mean square percentage error
 A measure of how much the modelpredicted values differ from the observed values of the series. It is independent of the units that are used and can therefore be used to compare series with different units.
 Root mean square error
 The square root of mean square error. A measure of how much a dependent series varies from its modelpredicted level, expressed in the same units as the dependent series.
 BIC
 Bayesian Information Criterion. A measure for selecting and comparing models based on the 2 reduced log likelihood. Smaller values indicate better models. The BIC also "penalizes" overparameterized models (complex models with a large number of inputs, for example), but more strictly than the AIC.
 AIC
 Akaike Information Criterion. A measure for selecting and comparing models based on the 2 reduced log likelihood. Smaller values indicate better models. The AIC "penalizes" overparameterized models (complex models with a large number of inputs, for example).
 Outliers over time
 Bar chart of the number of outliers, across all targets, for each time interval in the estimation period.
 Series transformations
 Table of any transformations that were applied to the series in the model system. The possible transformations are missing value imputation, aggregation, and distribution.