At the AnyLogic Conference 2023, we announced AnyLogic Cloud 2.3.3 and 2.4.0 releases and presented an optimization experiment that will be soon included in one of the upcoming updates.
Meanwhile, in this post, I will highlight the main features of the AnyLogic Cloud 2.3.3 release. Most of these are related to experiment dashboard charts:
- A scatter plot can be used to visualize model input-output dependency.
- A mean and error bar is available in variation with replications experiment.
- A time plot, time stack chart, time color chart, mean and error bar, and density plot now can be labeled with time or date.
- A model’s time unit is labeled with the respective time unit (e.g., seconds).
- A time plot, time stack chart, and time color chart support a non-default time unit.
- A box plot shows mean values.
- A time color chart has improved validation and color assignment to new items.
Let’s take a closer look at how these improvements can be used in simulation models.
A scatter plot for model input-output dependency
As you may know, in AnyLogic 8, there is a sensitivity analysis experiment type that performs a parameter variation with only one varied parameter. In AnyLogic Cloud, the analog is a variation experiment. Previously, Cloud only had a surface plot for plotting the input-output dependency. The surface was created in 3D, which was not convenient for observation of 2D data.
In Cloud 2.3.3 release, a scatter plot in variation experiment supports visualization of the input-output dependency. It provides the most convenient way to analyze the model's sensitivity to input values (or parameter variation) in 2D, visualizing non-aggregated data.
Let’s take the Consumer Credit example model and visualize the dependency of bank clerk utilization and the number of clerks with a scatter plot:
A mean and error bar available in a variation with replications experiment
This plot visualizes average output values along with a standard deviation. You can use it in Monte Carlo and variation experiments (from Cloud 2.3.3) where some model inputs are varied, and then all values from all iterations are aggregated and plotted, respectively. The mean and error bar is helpful when you want to estimate the fluctuation range of an output value.
A variation with replications experiment varies input values and runs each set of these values several times with a different random seed (similar to a Monte Carlo 2nd order experiment). The only difference is that in the variation with replications experiment, inputs are varied deterministically with a step value.
Now, with the mean and error bar, you can estimate the range using both stochastic and deterministic patterns of the input variation.
Plots labeled with time and date
Those who use both AnyLogic 8 and AnyLogic Cloud may have noticed that time plots (time plot, time stack chart, and time color chart) were not labeled with time units.
While in AnyLogic 8, model developers usually remember the model’s time unit, in AnyLogic Cloud, the models are executed by the final model user – a manager or an engineer. They are not necessarily familiar with model internals, such as the model’s time unit. A time unit used to be stated in the chart’s title:
It would be convenient if charts gave all the required information for data interpretation, such as a model’s time unit, without any extra actions from a user or developer. In AnyLogic Cloud 2.3.3, this feature is implemented – all time plots are labeled with the respective time units. You don’t need to ask a model developer about it or download a model’s open-source code to check it.
Besides, in the dashboard editor, you can always check what the current model’s time unit is and configure the chart to show data in a non-default time unit. This feature is helpful when you want to check, say, daily, weekly, and monthly reports using the same AnyLogic Cloud experiment dashboard.
Finally, a mean and error bar and a density plot can also be labeled with time units. These plots show aggregated data from several regular plots or time plots. Since you may combine the data from several sources with potentially different X axis labels, we have decided to let the user choose if they want to label the mean and error bar or density plot with a date/time label.
Mean value in a box plot
As a result of variation or Monte Carlo experiments, you receive aggregated statistics from all simple/simulation runs. This data includes min, max, median, and Q1/Q3 quartiles. However, a mean value is also an important metric, especially in stochastic model optimization, where an optimizer can choose the best set of inputs based on an average objective value.
We are sure a mean value is essential for aggregated data analysis, and in the Cloud 2.3.3 release, you can find it by hovering over a box plot or in a downloadable Excel file with experiment results.
An improved time color chart
Being similar to a Gantt chart, it’s used only to represent some object state, e.g., the busy or idle state of a resource. In the chart settings, you can map a specific state with a color, for example, busy – red, idle – green. Then use the chart to observe how the state changes over time.
The chart setup in AnyLogic Cloud used to be a bit tricky because each new mapping used the same color and condition, duplicated conditions were not validated, and so on.
From AnyLogic 2.3.3, each new mapping obtains a new color and condition, duplicated conditions are validated, and the interface is more user-friendly.
Custom branding for Private Cloud users
For Private Cloud administrators, we improved custom branding – now you can configure not only the color scheme and logo but also the title text, which appears in a browser tab. By default, the title suffix is "AnyLogic Cloud", but you can replace it with any text, for example, your company’s name.
All Cloud pages will show the new tab title suffix after you adjust the frontend.json configuration file and restart the respective services.
These are the main improvements in AnyLogic Cloud 2.3.3. We hope that you will enjoy working with the enhanced charts and will use them actively in your projects.
More details on all the improvements and updates released in Cloud 2.3.3 are in the release notes.
Stay tuned to our blog and subscribe to our monthly newsletter to make sure you don't miss any updates.