# Chart output

The types of output charts that are available depend on the measurement level of the target field, whether an iteration field is used, and the options that are selected in the Simulation Evaluation node dialog box. A number of the charts that are generated from a simulation have interactive features that you can use to customize the display. Interactive features are available by clicking Chart Options. All simulation charts are graphboard visualizations.

Probability density function charts for continuous targets. This chart shows both probability and frequency, with the probability scale on the left vertical axis and the frequency scale on the right vertical axis. The chart has two sliding vertical reference lines that divide the chart into separate regions. The table below the chart displays the percent of the distribution that is in each of the regions. If multiple density functions are displayed on the same chart (because of iterations), the table has a separate row for the probabilities that are associated with each density function, and an extra column that contains the iteration name and a color that is associated with each density function. The iterations are listed in the table in alphabetical order, according to the iteration label. If no iteration label is available, the iteration value is used instead. The table cannot be edited.

Each of the reference lines has a slider (inverted triangle) that you can use to easily move the line. Each slider has a label that indicates its current position. By default, the sliders are positioned at the 5th and 95th percentiles of the distribution. If there are multiple iterations, the sliders are positioned at the 5th and 95th percentiles of the first iteration that is listed in the table. You cannot move the lines to cross through each other.

A number of additional features are available by clicking Chart Options. In particular, you can explicitly set the positions of the sliders, add fixed reference lines, and change the chart view from a continuous curve to a histogram. See the topic Chart Options for more information. Right-click the chart to copy or export the chart.

Cumulative distribution function charts for continuous targets. This chart has the same two movable vertical reference lines and associated table that are described for the probability density function chart. The slider controls and table behave the same as the probability density function when there are multiple iterations. The same colors that are used to identify which density function belongs to each iteration are used for the distribution functions.

This chart also provides access to the Chart Options dialog box, which enables you to explicitly set the positions of the sliders, add fixed reference lines, and specify whether the cumulative distribution function is displayed as an increasing function (the default) or a decreasing function. See the topic Chart Options for more information. Right-click the chart to copy, export, or edit the chart. Selecting Edit opens the chart in a floating graphboard editor window.

Predicted category values chart for categorical targets. For categorical target fields, a bar chart displays the predicted values. The predicted values are displayed as the percent of the target field that is predicted to fall in each category. For categorical target fields with sensitivity analysis iterations, results for the predicted target category are displayed as a clustered bar chart that includes the results for all iterations. The chart is clustered by category or by iteration, depending on which option was selected in the Grouping for sensitivity analysis area in the Density Functions view of the Simulation Evaluation node dialog box. Right-click the chart to copy, export, or edit the chart. Selecting Edit opens the chart in a floating graphboard editor window.

Predicted category probabilities chart for categorical targets. For categorical target fields, a histogram displays the distribution of predicted probabilities for each of the categories of the target. For categorical target fields with sensitivity analysis iterations, the histograms are displayed by category or by iteration, depending on which option was selected in the Grouping for sensitivity analysis area in the Density Functions view of the Simulation Evaluation node dialog box. If the histograms are grouped by category, a drop-down list containing the iteration labels enables you to choose which iteration to display. You can also select the iteration to display by right-clicking the chart and selecting the iteration from the Iteration submenu. If the histograms are grouped by iteration, a drop-down list containing the category names enables you to choose which category to display. You can also select which category to display by right-clicking the chart and selecting the category from the Category submenu.

- Logistic, SVM, Bayes, Neural Net and KNN
- Db2/ISW in-database mining models for logistic regression, decision trees and naïve Bayes

Tornado charts. The tornado chart is a bar chart that shows the sensitivity of the target field to each of the specified inputs. The sensitivity is measured by the correlation of the target with each input. The title of the chart contains the name of the target field. Each bar on the chart represents the correlation between the target field and an input field. The simulated inputs that are included on the chart are the inputs that are selected in the Input fields for correlation tornado field on the Fields view of the Simulation Evaluation node dialog box. Each bar is labeled with the correlation value. Bars are ordered by the absolute value of the correlations, from largest value to smallest. If there are iterations, a separate chart is generated for each iteration. Each chart has a subtitle, which contains the name of the iteration.

Distribution table. This table contains the value of the target field, below which the specified percent of the observations fall. The table contains a row for each percentile value that is specified on the Outputs view of the Simulation Evaluation node dialog box. The percentile values can be quartiles, a different number of equally spaced percentiles, or individually specified percentiles. The distribution table contains a column for each iteration.

Information. This section gives an overall summary of the fields and records that are used in the evaluation. It also shows the input fields and record counts, which are broken down for each iteration.