Scenario Chart

Overview and Key Concepts

The Scenario Chart allows you to compare data from another chart across multiple Experiment scenarios (or Optimizer solutions) on a single chart.

To use the Scenario Chart, you must specify a target chart. The Scenario Chart will mimic the type of the target chart; if the target chart is a Bar Chart, the Scenario Chart will also be a Bar Chart. The Scenario Chart is compatible any of the following chart types:

In addition to specifying a target chart, you must also specify which scenarios you want to compare. The Scenario Chart uses this information to produce a chart that looks similar to the original chart, and also compares the requested scenarios. Since the Experimenter usually runs multiple replications per scenario, the Scenario Chart will aggregate data from all replications of each scenario. This way, the Scenario Chart will form a fair visual representation of the target chart's information for each scenario.

Replication Aggregation

The Scenario Chart will aggregate data from all replications of each chosen scenario. Depending on the target chart's type, data may be aggregated quite differently. The following sections describe how the aggregation works in general.

Table Charts

If the Scenario Chart is based on a table chart, you will need to specify which columns of the table to aggregate, and how they should be aggregated. All other columns are considered identifier columns, and are used to determine which rows should be aggregated together.

For example, suppose you had a table chart that showed the content of several queues. Then suppose you ran an experiment with two scenarios, and three replications per scenario. Each replication would produce a table chart, like the ones shown in the following image:

If you chose to compare Scenario 1 and Scenario 2, and to aggregate the WIP column, the Scenario Chart would look like the following:

Pie and Bar Charts

If the Scenario Chart is based on a Pie or Bar chart, then the Scenario chart will calculate the average value of each bar or pie segment across all replications of each scenario. Those averaged values will then be used to create an average version of each bar group or pie ring for each scenario.

For both charts, you may need to specify whether the text shown on the chart needs to be aggregated also. For example, a state pie chart might show a utilization percentage in the center. If text is present on each bar group or pie ring, but that text is not aggregated, then the Scenario Chart will use that text to identify which bar or segment values should be grouped together for aggregation.

Box Plots and Histograms

If the Scenario Chart is based on a Box Plot or Histogram, then the resulting chart will be a box plot or histogram of all data from all replications, for each scenario.

Time Plots

If the Scenario Chart is based on a Time Plot, you can specify an interpolation mode. If the interpolation mode is None, then all data from all replications will be drawn on the plot, colored by Scenario. Otherwise, you must addtionally specify a sample rate and a confidence interval. The Scenario Chart will create time intervals, based on the sample rate, and sample the data at those intervals. For each replication, it is likely that a data point was not recorded at the interval time, so the Scenario Chart will interpolate to find the value. Those interpolated values are then aggregated to determine the mean. In addition, you can also show the min, max and confidence interval values.

Properties Panels

The scenario chart uses the following properties panels:

Other panels will appear in the properties window. However, the other panels deal with the settings specific to the specified chart.