Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
The BayesiaLab WebSimulator is a web application that allows you to share interactive models with your internal or external audience (e.g., your research clients), without having to install any software on their computers and without having end-users be familiar with the concept of Bayesian networks.
The BayesiaLab WebSimulator publishes an interactive web page based on an XBL file, a Bayesian network model you created in BayesiaLab.
You can designate which of the nodes in your network should be displayed to the end-user as Inputs and Outputs.
Upon publishing the network, the BayesiaLab WebSimulator utilizes the Bayesia Engine API — entirely in the background — to perform inference in your network. This allows end-users to enter observations/evidence on the nodes in your network via their web browser. Whenever evidence is set, the user will immediately see updated probability distributions of the Outputs.
The BayesiaLab WebSimulator's web interface is responsive, allowing even tablet or smartphone users to work dynamically with your network.
https://simulator.bayesialab.com/#!questionnaire/377786492943
The WebSimulator infrastructure is comprised of three main constituents:
The analyst/developer uses BayesiaLab on his local hardware to build and configure a Bayesian network model for publication.
The BayesiaLab WebSimulator Server, located on the premises of Bayesia S.A.S., hosts and serves the uploaded and published model.
The end-users access the published model using a web browser. Importantly, models can be made available in two ways:
To a restricted group of users with a Private WebSimulator Account.
To the general public through a Public WebSimulator Account.
The overarching concept of "WebSimulator" actually covers two closely related applications, the WebSimulator proper and the Adaptive Questionnaire.
Both the WebSimulator and the Adaptive Questionnaires allow end-users to perform interactive inference via a web interface.
In a WebSimulator, the end-user sets observations on the Inputs, which updates the posterior probability distributions of the Outputs.
In an Adaptive Questionnaire, as the end-user sets observations, the remaining Inputs are dynamically ordered, from high to low, according to the information they bring to the Target (or the set of Targets), given the current set of observations. The Adaptive Questionnaire provides a dynamic recommendation for seeking the optimal next piece of evidence with the objective of reducing the uncertainty of the Targets.
Whenever we refer to "WebSimulators" in general, we mean both types of simulators available in BayesiaLab: the WebSimulator proper and the Adaptive Questionnaire.
You must have already created a Bayesian network model in BayesiaLab to publish a WebSimulator or Adaptive Questionnaire.
The following two steps explain your workflow for configuring and publishing WebSimulators and Adaptive Questionnaires:
Preparing Your Model — In the WebSimulator Editor in BayesiaLab, you configure the functionality and appearance of your to-be-published model.
Publishing Your Model — The WebSimulator Administration Page (i.e., the web interface of the WebSimulator server) allows you to upload, apply modifications, and publish your model.
Please note that the process for creating a WebSimulator and an Adaptive Questionnaire is nearly identical.
This section applies to end-users, i.e., the individuals who will access the WebSimulator via web browser and utilize the published model for interactive inference.
There are no software or hardware requirements for end-users other than a web browser and an Internet connection.
List of Available Nodes in the current Bayesian network that can be used in the WebSimulator as Inputs or Outputs.
This list shows the Available Metrics that can be used as Outputs in the WebSimulator of the network.
The panels for Available Nodes and Available Metrics appear identically in all three tabs of the WebSimulator Editor, i.e., Simulator, Inputs, and Outputs.
This panel displays the nodes that are selected to be published as Inputs in the WebSimulator.
States
Probabilities
Mean
Mean Mode
Binary
MinXEnt
Displayed Name specifies the name of the component in the WebSimulator. By default, it is the name of the node or the metric.
Clicking the Long Name button adopts any Long Name that you may have defined for the node when editing the network in BayesiaLab.
You can also edit the Displayed Name.
Font Color specifies the color of the Displayed Name in the WebSimulator.
Image Size — The image displayed in the component is square, and you can specify the length of the square's sides in pixels.
Number Output Format
Clicking on Number Output Format, you can specify the number format of the Input. Note that this option is only available for Probabilities-type and Mean-type Inputs.
The Description field provides you with space for additional information regarding the output component.
The published WebSimulator displays this content when end-users hover with their cursor over the component.
By default, Description adopts the Comment you may have added to the node while editing the network in BayesiaLab.
This list displays the Available Nodes in the current Bayesian network that can be used as Inputs or Outputs in the WebSimulator of the network.
This list shows the Available Metrics that can be used as Output Components in the WebSimulator of the network.
The panels for Available Nodes and Available Metrics appear identically in all three tabs of the WebSimulator Editor, i.e., Simulator, Inputs, and Outputs.
Name of the WebSimulator, as it will appear in the list of the models.
You can specify the Title to be shown at the top of the published WebSimulator page.
Name of the WebSimulator's Author.
For public WebSimulators, indicating the Field of Study can help your audience find your model on the WebSimulator Home Page.
For public WebSimulators, indicating the Type of Analysis can help your audience find your model on the WebSimulator Home Page.
Theme Color refers to the background color of the horizontal panel at the top of your published WebSimulator.
By default, the Description field adopts the content of the Comment associated with your Bayesian network model.
For public WebSimulators, this Description also appears on the WebSimulator Homepage when end-users hover with their cursor over the preview tile.
This picture is also used for your model's preview tile on the WebSimulator homepage.
You can specify a Logo Picture or banner to appear in the upper left corner of your model's WebSimulator page. Most screenshots on this topic page feature the banner of the 2018 BayesiaLab Conference in Chicago.
Logo URL allows you to associate a hyperlink with the Logo Picture.
Font Size allows you to specify the size of the font displayed on your model's WebSimulator page.
Input Layout refers to the style in which the Input elements appear in the Input Panel.
History Layout is only applicable to Adaptive Questionnaires. This option defines how the pieces of evidence that the end-user has set appear in the Input Panel.
Output Layout refers to the style in which the Output elements appear in the Output Panel.
If you check the Show Variations option, your model's WebSimulator will highlight the impact of the pieces of evidence set by the end-user.
If you check the Reference State option, the WebSimulator will display variations relative to the marginal distributions of the Outputs.
If you specify Targets by dragging available nodes into the Targets panel, your WebSimulator will serve as an Adaptive Questionnaire.
Unlike the Adaptive Questionnaire function in BayesiaLab (Main Menu > Inference > Adaptive Questionnaire
), an Adaptive Questionnaire in the WebSimulator permits multiple Targets.
For Input Layout, History Layout, and Output Layout, you can choose from the following styles:
The process of publishing an existing Bayesian network model starts with the WebSimulator Editor in BayesiaLab.
The WebSimulator Editor allows you to configure the to-be-published model within BayesiaLab.
In BayesiaLab, you open the WebSimulator Editor by selecting Main Menu > Tools > WebSimulator Editor
.
To configure a WebSimulator or an Adaptive Questionnaire, you need to go through each tab of the WebSimulator Editor as described in the following sections:
The configurations of each tab are automatically associated with your Bayesian network model.
As you save your model, these settings are automatically recorded in the same XBL file.
Whenever you reopen your Bayesian network model, you can return to the WebSimulator settings and modify them as necessary.
Once you have saved your fully configured model as an XBL file, you can leave BayesiaLab and proceed to the WebSimulator Administration Page to upload and publish your model.
Available Nodes
Available Metrics
Inputs
You can add nodes to the list by dragging them from the list of Available Nodes to the list of Inputs .
Upon adding a node to the list of Inputs , you will be prompted to select the format of the Input:
Switches
Combo Box
Slider
Text
Slider
Text
Both
Displayed Name
Font Color
Component Image
Component Image allows you to select an image or icon to be displayed with the component. By default, any image that is already associated with the node in BayesiaLab will be used.
Image Size
Description
Available Nodes
Available Metrics
Name
Title
Author
Field of Study
Type of Analysis
Theme Color
Description
Description allows you to provide a description or commentary, which is available to the end-user by clicking on the Information icon on your WebSimulator page.
Description Picture
With Description Picture, you can specify a pictorial representation of your model. You can upload an image file of your choice or click the snapshot icon to capture a picture of your current Bayesian network.
This image is available to the end-user when clicking on the Eye icon on the WebSimulator page.
Logo Picture
Logo URL
Font Size
Input Layout
History Layout
Output Layout
Show Variations
Reference State
Targets
Grid
Flow
Line
Accordion
Once you save these settings with your Bayesian network file, you can upload and publish it via the .
The WebSimulator Editor is available both in the and the .
, to define the general settings of the WebSimulator.
to define the interface for end-user input.
specify how the outputs will be shown to the end-user.
BayesiaLab indicates the presence of a WebSimulator configuration with the WebSimulator indicator icon in the bottom-right corner of the main window.
This list displays the Available Nodes in the current Bayesian network that can be used as Inputs or Outputs in the WebSimulator of the network.
This list shows the Available Metrics that can be used as Outputs in the WebSimulator of the network.
The panels for Available Nodes and Available Metrics appear identically in all three tabs of the WebSimulator Editor, i.e., Simulator, Inputs, and Outputs.
This panel displays the nodes that are selected to be published as Outputs in the WebSimulator.
To add nodes or metrics as Output Components, you select the nodes or metrics on the Available Nodes or Available Metrics lists and then drag and drop them onto the Outputs panel.
You can remove an item from the Outputs panel by dragging it back to its origin.
You can reorder the list items by dragging them to the desired positions on the list. The specified order will be retained in the WebSimulator display.
As you drop nodes or metrics onto the Outputs list, you will be prompted to select the appearance of their corresponding components in the WebSimulator. The options depend on the type of node or metric:
Probabilistic Node — Display Options:
Mean Bar
Mean Gauge
Probability Bar
Probability Text
Utility Node — Display Options:
Utility Bar
Utility Gauge
Utility Text
Joint Probability Metric — Display Options:
Joint Probability Bar
Joint Probability Gauge
Joint Probability Text
Global Utility Metric — Display Options:
Global Utility Bar
Global Utility Gauge
Global Utility Text
Decision Node:
Quality Bar
Function Node
Function
The Displayed Name specifies the component's name label in the WebSimulator. By default, it is the name of the node or the metric.
Clicking the Long Name button adopts any Long Name that you may have defined for the node when editing the network in BayesiaLab.
You can also edit the Displayed Name.
Font Color specifies the color of the Displayed Name in the WebSimulator.
Component Image allows you to select an image or icon to display with the component. By default, any image already associated with the node in BayesiaLab will be used.
Image Size — The image displayed in the component is square, and you can specify the length of the square's sides in pixels.
Output Format for Function Node
If a Function Node is selected as an Output component, a separate Output Format button allows you to set the number format separately from the other components. By default, the Function Node component adopts the number format defined in the Properties tab of the Function Node Editor.
The Description field provides you with space for additional information regarding the output component. The published WebSimulator displays this content when end-users hover with their cursor over the component. By default, Description adopts the Comment you may have added to the node while editing the network in BayesiaLab.
Preview
The Preview button opens up a static preview of the WebSimulator in your default web browser to let you see its appearance before publishing it. This lets you quickly fine-tune the layout and experiment with various component settings.
The Preview is static and does not perform any simulation. While you can experiment with the inputs, the output side of the WebSimulator will not update.
Reset Button
The WebSimulator Administration Page is your interface to the BayesiaLab WebSimulator Server, which is physically hosted on the premises of Bayesia S.A.S.
On the Administration Page, you can publish a WebSimulator that you designed and configured in BayesiaLab through the WebSimulator Editor.
Open a browser and go to https://simulator.bayesialab.com.
Alternatively, you can proceed directly to http://simulator.bayesialab.com/#!adminlogin
Enter your Username and Password and click Login.
Your Username and Password are the same as the credentials that you use to log in to the Bayesia License Server (BLS).
If you are a user of a Single-User/Single-Machine License (SUSM), you may not yet have credentials to access BLS.
In that case, you need to use your Client Identifier (format: AAAA-BBBB-CCCC-DDDD) to sign in to the Bayesia License Server and register. Please follow the steps explained on the BLS Login page.
If your organization has multiple accounts with Bayesia, you need to ensure that the account you use has the appropriate privileges for publishing public or private models.
These subsections explain the core processes of uploading, publishing, and managing your previously configured Bayesian network model.
Although they are both very similar, we present the workflows for WebSimulator and Adaptive Questionnaire separately for clarity in terminology:
Publishing a WebSimulator
Publishing an Adaptive Questionnaire
Either workflow concludes the publication process and makes your model available for use by the intended audience.
You may wish to refer your audience to the end-user guides:
Using the WebSimulator
Using the Adaptive Questionnaire
This section covers the process of publishing a fully configured Bayesian network model as a WebSimulator.
You need to have a Bayesian network model available that was configured with the WebSimulator Editor in BayesiaLab.
You have successfully logged in to the WebSimulator Administration Page.
Click the New Simulator button in the upper right corner of the Administration Page.
A new pop-up window presents you with four tabs:
Bayesian Network
Description
Properties
Title Bar
Click the Select File button to choose the Bayesian network (in XBL format) that you have configured for use with the WebSimulator.
Click OK to upload the selected file to the server.
If the uploaded file meets the requirements, the message "Valid" appears behind the Select File button.
Many of the settings in this tab and the following tabs repeat options that you could have specified in the WebSimulator Editor before uploading the XBL file to the WebSimulator Server. As a result, many of the fields contain default values. However, you can modify any existing content by providing new values. Please note that such edits only apply "downstream" to the WebSimulator but will not propagate backward to your Bayesian network file.
By default, the Name field adopts any name you specified in the Simulator tab of the WebSimulator Editor. Note that this field is mandatory.
The slide-in message on the bottom-right corner of the browser window states that all observations have been removed from your model. This means that any dataset that had been associated with your XBL was immediately stripped and is no longer available on the WebSimulator Server.
The Description tab allows you to provide background and contextual information regarding your Bayesian network model.
By default, this field adopts the Description defined in the Simulator tab of the WebSimulator Editor, which, in turn, utilizes any available Network Comments associated with the Bayesian network model.
Furthermore, you can remove any existing Image and replace it with a visual of your choice by clicking on Select File.
Field of Study and Analysis Type allow you to characterize your Bayesian network model. This can help you direct end-users to your model, especially with public WebSimulators.
The elements of the Properties tab repeat some of the options that are also available in the Simulator tab of the WebSimulator Editor.
If you check the Display Variations option, your model's WebSimulator will highlight the impact of the pieces of evidence set by the end-user.
If you check the Store Initial Reference State option, the WebSimulator will display variations relative to the marginal distributions of the Outputs.
For an illustration of the Layout Options, please see Layout in the WebSimulator Editor — Simulator Settings.
You can specify a Logo or banner to appear in the upper left corner of your model's WebSimulator page.
Logo URL allows you to associate a hyperlink with the Logo.
You can specify the Title to be shown at the top of the published WebSimulator page.
Theme Color refers to the background color of the horizontal panel at the top of your published WebSimulator.
Once your WebSimulator is set up, you can proceed to publish it.
From the Administration Page, select your model and then click the Publish button.
Depending on your subscription type, you may only have the public option available or both the public and the private.
The dates show the respective expiration dates of your subscriptions.
"Public" means that anyone with an Internet connection can access the WebSimulator of your published model.
A public WebSimulator will also appear as a tile on the public WebSimulator homepage at https://simulator.bayesia.com. This is a great way to showcase your work to the wider world.
By default, all BayesiaLab Professional licenses allow you to publish public WebSimulators.
"Private" means that you can restrict access to your WebSimulator to users whom you have given a specific URL plus a password.
To publish private and password-protected models, you need to subscribe to a private WebSimulator Account. Please see the pricing guide for more details.
A private WebSimulator Account allows you to publish both private WebSimulators and private Adaptive Questionnaires.
As you publish a private model, you will be prompted to create a password that will be linked to the URL of your WebSimulator.
The password should have at least 8 characters, including at least one number.
This means that access to your private WebSimulator is controlled by one URL and one password.
Anyone to whom you give the URL plus the password will have access to your private WebSimulator.
To revoke access to your private WebSimulator, you need to unpublish it.
Even once you have subscribed to a private WebSimulator account, you still retain your free public account, so you have both options available when publishing a model.
Be careful not to inadvertently publish a model publicly that you intended to share privately.
The Show Details button on the Administrator Page allows you to view all the WebSimulator settings highlighted in the list.
A pop-up window shows all settings specified in the New Simulator Workflow and also features the same tabs:
Bayesian Network
Description
Properties
Title Bar
The Bayesian Network tab, however, provides additional elements:
The Identifier is an automatically assigned 12-digit number that uniquely identifies your WebSimulator.
Creation Date
Modification Date
Published Date
Password reveals the password you set for the WebSimulator when you published it. For public WebSimulators, this field is blank.
Direct URL is the URL you can share with end-users so they can go directly to your WebSimulator.
The format of the URL follows the pattern: https://simulator.bayesialab.com/#!simulator/Identifier
Clicking the Edit Properties button allows you to modify the properties that you previously specified in the New Simulator Workflow or, before that, in the WebSimulator Editor.
Edit Properties is only available for unpublished WebSimulators. If you wish to modify a currently published WebSimulator, you must first click the Unpublish button on the Administration Page.
Upon modification, you can click Publish to republish your WebSimulator again.
Clicking the Run button opens up your WebSimulator so you can see it and use it as an end-user.
More specifically, your browser opens the URL https://simulator.bayesialab.com/#!simulator/Identifier
This is now the live version of the WebSimulator, of which you could previously only launch a static preview in the WebSimulator Editor.
This section covers the process of publishing a fully configured Bayesian network model as an Adaptive Questionnaire.
Click the New Questionnaire button in the upper right corner of the Administration Page.
A new pop-up window presents you with four tabs:
Bayesian Network
Description
Properties
Title Bar
Click the Select File button to choose the Bayesian network (in XBL format) that you have configured for use with the Adaptive Questionnaire.
Click OK to upload the selected file to the server.
If the uploaded file meets the requirements, the message "Valid" appears behind the Select File button.
The slide-in message on the bottom-right corner of the browser window states that all observations have been removed from your model. This means that any dataset that had been associated with your XBL was immediately stripped and is no longer available on the WebSimulator Server.
The Description tab allows you to provide background and contextual information regarding your Bayesian network model.
By default, this field adopts the Description defined in the Simulator tab of the WebSimulator Editor, which, in turn, utilizes any available Network Comments associated with the Bayesian network model.
Furthermore, you can remove any existing Image and replace it with a visual of your choice by clicking on Select File.
Field of Study and Analysis Type allow you to characterize your Bayesian network model. Especially with public WebSimulators, it can help you direct end-users to your model.
If you check the Display Variations option, the Questionnaire will highlight the impact of the pieces of evidence set by the end-user, either with respect to the previous state or the Reference State (see next item).
If you check the Store Initial Reference State option, the Questionnaire will display variations relative to the marginal distributions of the Outputs. More specifically, the Initial Reference State refers to the probabilities and values of the marginal distribution of the Outputs before setting any observations. With this option checked, the variations will be computed as the difference between the current state and the reference state. Otherwise, the variations are computed as the difference between the current state and the previous state, i.e., before setting the most recent piece of evidence.
You can specify a Logo or banner to appear in the upper left corner of your model's Questionnaire page.
Logo URL allows you to associate a hyperlink with the Logo.
You can specify the Title to be shown at the top of the published WebSimulator page.
Theme Color refers to the background color of the horizontal panel at the top of your published Questionnaire.
Once your Questionnaire is set up, you can proceed to publish it.
From the Administration Page, select your model and then click the Publish button.
Depending on your subscription type, you may only have the public option available or both the public and the private.
"Public" means that anyone with an Internet connection can access the Questionnaire of the model you published.
By default, all BayesiaLab Professional licenses allow you to publish public Questionnaires.
"Private" means that you can restrict access to your Questionnaire to users whom you have given a specific URL plus a password.
To publish private and password-protected models, you need to subscribe to a private WebSimulator Account. Please see the pricing guide for more details.
A private WebSimulator Account allows you to publish both private WebSimulators and private Adaptive Questionnaires.
As you publish a private model, you will be prompted to create a password that will be linked to the URL of your Questionnaire.
The password should have at least 8 characters, including at least one number.
This means that access to your private Questionnaire is controlled by one URL and one password.
Anyone to whom you give the URL plus the password will have access to your private Questionnaire.
To revoke access to your private Questionnaire, you need to unpublish it.
Even once you have subscribed to a private WebSimulator Account, you still retain your free public account and thus have both options available when publishing a model.
Be careful not to inadvertently publish a model publicly that you intended to share privately.
The Show Details button on the Administrator Page allows you to view all settings of the WebSimulator highlighted in the list.
Bayesian Network
Description
Properties
Title Bar
The Bayesian Network tab, however, provides additional elements:
The Identifier is an automatically assigned 12-digit number that uniquely identifies your Questionnaire.
Creation Date
Modification Date
Published Date
Password reveals the password you set for the Questionnaire when you published it. For public Questionnaires, this field is blank.
Direct URL is the URL you can share with end-users so they can go directly to your Questionnaire.
The format of the URL follows the pattern: https://simulator.bayesialab.com/#!simulator/Identifier
Edit Properties is only available for unpublished Questionnaires. If you wish to modify a currently published Questionnaire, you first need to click the Unpublish button on the Administration Page.
Upon modification, you can click Publish to republish your Questionnaire again.
Clicking the Run button opens up your Questionnaire so you can see it and use it like an end-user.
More specifically, your browser opens the URL https://simulator.bayesialab.com/#!simulator/Identifier
For end-users, there are three ways to reach the page of an Adaptive Questionnaire:
Only models that were published using a public account will appear in the drop-down menu.
Select any available models from the drop-down menu to open it.
You open a specific URL in the format https://simulator.bayesialab.com/#!questionnaire/Identifier
and directly go to that particular Adaptive Questionnaire page.
A public Adaptive Questionnaire page opens immediately.
A private Adaptive Questionnaire page prompts you for a password.
Upon opening the page of the selected model, you see the layout specified by the designer of the Adaptive Questionnaire.
The Adaptive Questionnaire features three panels:
The Open Questions Panel (top left) shows the unobserved variables according to their importance with regard to the Targets.
The Answered Questions Panel (bottom left) lists all the variables that have been observed or intentionally skipped.
The Output Panel (right) features the Targets, i.e., the variables of interest that we wish to predict.
At the top of the Open Questions Panel, a drop-down menu offers you other available Adaptive Questionnaires to which you can switch at any time by clicking on the corresponding list item.
Selecting the first (blank) item in the list closes your current Adaptive Questionnaire and loads a blank page.
To the right of the Adaptive Questionnaire Selector, there are two icons:
The order of the Inputs in the panel — left-to-right, top-to-bottom — indicates the optimal sequence in which the "open questions" should be answered.
More specifically, the information gain determines the sort order — from high to low — each Input brings towards the Targets.
You enter an observation, i.e., a piece of evidence, by changing the state, the value, or the slider position of an Input.
Immediately upon entering your evidence, the corresponding Input is marked Observed and moved to the Answered Questions Panel below.
The evidence is applied to the corresponding node in the underlying Bayesian network, which computes the posterior probability distributions of Inputs and Targets. Their updated distributions are displayed instantly.
Furthermore, the information gain of the Inputs is recalculated, and their order is updated dynamically in the Open Questions Panel.
Skipping a question will move the corresponding Input to the Answered Questions Panel below.
Note that despite the recommended order, you can still set evidence on any Inputs in any order.
Ordering the Inputs also takes into account any so-called Costs of observing a variable.
In BayesiaLab, Cost captures the "price to pay" to obtain an observation. In the medical domain, for instance, measuring the blood pressure of a patient is presumably less expensive than performing an electrocardiogram. Cost could refer to an actual monetary amount of observing a specific Input or merely express a relative cost, e.g., it requires twice as much effort/time/resources to observe X than to observe Y.
All Costs are defined in the underlying Bayesian network and cannot be modified by the end-user of the Adaptive Questionnaire.
Once an Input has been observed and moved to the Answered Questions panel, you can still modify the evidence on that Input.
Alternatively, you can remove an observation altogether by unchecking the Observed box on the Input. In that case, the input moves back to the Open Questions Panel.
As a result, all Inputs move back up to the Open Questions Panel.
Now, the Outputs display the changes that resulted from setting the most recent piece of evidence.
The deltas are shown in green and red for positive and negative changes respectively.
By default, the variations are calculated with respect to the previous state, i.e., before setting the most recent observation.
This shows the incremental change brought to the Outputs by each piece of evidence.
However, if you are interested in the impact of a set of multiple observations, you first need to set a reference state:
The deltas shown on the Outputs now are calculated based on the Reference State.
Observations Sets allow you to save sets of evidence so you can retrieve and review them later, without having to enter each observation again individually.
You can also share your saved Observation Sets with other end-users for their review.
Furthermore, BayesiaLab users can save their Observation Sets, such as the best solutions from an optimization performed with BayesiaLab, and share them with Adaptive Questionnaire users so they can review the Observation Sets that are proposed as optimal solutions.
At any time in the Adaptive Questionnaire, you can save the state of observations as an Observation Sets. This even includes non-evidence, i.e., Inputs that have not been observed. Thus, your Observation Sets can include observed and not-observed Inputs.
A pop-up window prompts you to specify a name for the to-be-stored Observation Set.
Enter the name and click OK.
The new Observation Set is like a running tab to which you can add further observations under the same Observation Set Name.
If you select an existing Observation Set Name, the evidence you save will be sequentially numbered, starting with 0.
Click on the Observation Set you wish to restore, and the Adaptive Questionnaire sets all observations accordingly and performs inference on that basis.
It is important to realize that the Adaptive Questionnaire is not merely returning saved results. Instead, the Adaptive Questionnaire only retrieves the saved Inputs and then performs inference again on that basis. Hence, all results shown are freshly recalculated.
You can clear all Observation Sets that are currently stored in your Adaptive Questionnaire session.
By default, your stored Observations Sets only remain available during your current Adaptive Questionnaire session. If you closed your browser and reopened the same Adaptive Questionnaire again, all Observation Sets would be gone.
To save Observation Sets permanently, you can export them to a Scenario File.
Once you confirm, your browser saves a file named observations.txt
to your local drive.
This file can be shared with other WebSimulator and BayesiaLab users so they can retrieve and replicate your saved Observation Sets.
The format of this Scenario File is identical to the Evidence Scenario File produced by BayesiaLab. As such, it is an ideal format for exchanging scenarios for review and discussion between BayesiaLab users and WebSimulator users.
Similarly, you can load a locally saved Scenario File back into the Adaptive Questionnaire.
Please note the important distinction between temporarily storing an Observation Set for the duration of your WebSimulator session and permanently saving an Observation Set to an external Scenario File on a local drive.
Whenever you have stored Observation Sets available in your Adaptive Questionnaire — indicated by the Observation Set icon — you can enter the so-called Simulation Mode. This mode allows you to easily scroll through all available Observation Sets.
You can now step through all Observation Sets backward and forward.
The name of the current Observation Set is shown on a slide-in label in the bottom right corner of your Adaptive Questionnaire window.
You can further modify any Observation Set retrieved in that fashion; however, you can't save that modified evidence as a new Observation Set. To do so, you will first need to exit the Simulation Mode.
Beyond the default layout specified by the publisher of the Adaptive Questionnaire, you can choose your own layout to optimize the Adaptive Questionnaire of your screen size, resolution, and format.
You can choose the layouts separately for the Input and the Output panel, but the options are the same for both.
Input refers to the Open Questions Panel.
History refers to the Answered Questions Panel.
Output refers to the Output Panel.
For end-users, there are three ways to reach a WebSimulator page:
In the header, you can filter by Field of Study, Analysis Type, and Model Type.
Only models that were published using a public account will appear here.
Click on any of the available models to open them.
You open a specific URL in the format https://simulator.bayesialab.com/#!simulator/Identifier
and directly go to that particular WebSimulator page.
A public WebSimulator page opens immediately.
A private WebSimulator page prompts you for a password.
Upon opening the page of the selected model, you see the layout specified by the designer of the WebSimulator.
The WebSimulator features two panels:
Input Panel (left)
Output Panel (right)
At the top of the Input Panel, a drop-down menu offers you other available WebSimulators, which you can switch to at any time by clicking on the corresponding list item.
Selecting the first (blank) item in the list closes your current WebSimulator and loads a blank page.
To the right of WebSimulator Selector, there are two icons:
Please note that we use "observation" and "piece of evidence" as well as "setting evidence," "setting observations," and "observing" interchangeably.
Depending on the type of Input, you can set an observation in a number of ways:
Upon setting evidence on an Input, the Observed checkbox under the corresponding Input is checked.
With that, this new evidence is set to the node in the underlying Bayesian network, which computes the posterior probability distributions of all other nodes in the network.
As a result, the WebSimulator displays updated Inputs and Outputs.
Note that a Bayesian network always performs omnidirectional inference, so new evidence can affect all nodes, not just the Outputs (see What is a Bayesian Network? Probabilistic Inference)
There are two ways to remove any evidence you have set:
Alternatively, you can remove the evidence set on an individual Input by unchecking the Observed box.
Now, the Outputs display the changes that resulted from setting the most recent piece of evidence.
The deltas are shown in green and red for positive and negative changes respectively.
By default, the variations are calculated with respect to the previous state, i.e., before setting the most recent observation.
This shows the incremental change brought to the Outputs by each piece of evidence.
However, if you are interested in the impact of a set of multiple observations, you first need to set a reference state:
Select Store Reference State
.
Observations Sets allow you to save sets of evidence to retrieve and review them later without having to enter each observation again individually.
You can also share your saved Observation Sets with other end-users for their review.
Furthermore, BayesiaLab users can save their Observation Sets, such as the best solutions from an optimization performed with BayesiaLab, and share them with WebSimulator users so they can review the Observation Sets that are proposed as optimal solutions.
At any time in the WebSimulator, you can save the state of observations as an Observation Set. This even includes non-evidence, i.e., Inputs that have not been observed. Thus, your Observation Sets can include observed and not-observed Inputs.
A pop-up window prompts you to specify a name for the to-be-stored Observation Set.
Enter the name and click OK.
The new Observation Set is like a running tab to which you can add further observations under the same Observation Set Name.
If you select an existing Observation Set Name, the evidence you save will be sequentially numbered, starting with 0.
Click on the Observation Set you wish to restore, and the WebSimulator sets all observations accordingly and performs inference on that basis.
It is essential to realize that the WebSimulator is not merely returning saved results. Instead, the WebSimulator only retrieves the saved Inputs and then performs inference again on that basis. Hence, all results shown are freshly recalculated.
You can clear all Observation Sets that are currently stored in your WebSimulator session.
By default, your stored Observations Sets only remain available during your current WebSimulator session. If you closed your browser and reopened the same WebSimulator again, all Observation Sets would be gone.
To save Observation Sets permanently, you can export them to a Scenario File.
Once you confirm, your browser saves a file named observations.txt
to your local drive.
This file can be shared with other WebSimulator and BayesiaLab users so they can retrieve and replicate your saved Observation Sets.
Similarly, you can load a locally saved Scenario File back into the WebSimulator.
Please note the important distinction between temporarily storing an Observation Set for the duration of your WebSimulator session and permanently saving an Observation Set to an external Scenario File on a local drive.
Whenever you have stored Observation Sets available in your WebSimulator — indicated by the Observation Set icon — you can enter the so-called Simulation Mode. This mode allows you to scroll through all available Observation Sets easily.
You can now step through all Observation Sets backward and forward.
The name of the current Observation Set is shown on a slide-in label in the bottom right corner of your WebSimulator window.
You can further modify any Observation Set retrieved in that fashion; however, you can't save that modified evidence as a new Observation Set. To do so, you will first need to exit the Simulation Mode.
Beyond the default layout specified by the publisher of the WebSimulator, you can choose your own layout to optimize the WebSimulator of your screen size, resolution, and format.
You can choose the layouts separately for the Input and the Output panel, but the options are the same for both.
Available Nodes
Available Metrics
Outputs
Displayed Name
Font Color
Component Image
Image Size
Number Output Format
Clicking on Number Output Format, you can specify the number format of all components, with the exception of the Function Nodes.
Description
A reset button is available for most settings. Click it to return to the default value.
Click on the navigation icon and then select Go to Admin.
Clicking the Account icon in the upper right corner of the Administration Page brings up a drop-down menu that allows you to:
Change Account , i.e., select another account associated with your organization.
Log Out from the Administration Page.
The same applies to the Author field. Note that the Author information will appear in the context of the properties of your model, which are available to the end-user by clicking the information icon in the published WebSimulator.
End-users can access the Description by clicking the Information icon in the published WebSimulator.
If an Image was already provided in the WebSimulator Editor, the eye icon will indicate its presence. Typically, you would use an Image depicting the Bayesian network model. Clicking the eye icon provides you with a preview.
Once you have published your model, its symbol will switch from to in the Published column of the Administration Page.
The Sharing column indicates whether you have shared your model publicly or privately .
You can generally publish your Bayesian network model as a public or a private WebSimulator.
You need to have a Bayesian network model available that was configured with the in BayesiaLab.
You have successfully logged in to the .
Many of the settings in this tab and the following tabs repeat options that you could have specified in the before uploading the XBL file. As a result, many of the fields contain default values. However, you can modify any existing content by providing new values. Please note that such edits only apply "downstream" to the WebSimulator, but they will not propagate backward to your Bayesian network file.
By default, the Name field adopts any name you specified in the tab of the WebSimulator Editor. Note that this field is mandatory.
The same applies to the Author field. Note that the Author information will appear in the context of the properties of your model, which are available to the end-user by clicking the information icon in the published WebSimulator.
End-users can access the Description by clicking the Information icon in the published WebSimulator.
If an Image was already provided in the WebSimulator Editor, the eye icon will indicate its presence. Typically, you would use an Image depicting the Bayesian network model. Clicking the eye icon provides you with a preview.
The elements of the Properties tab repeat some of the options that are also available in the Simulator tab of the .
For an illustration of the Layout Options, please see Layout in the .
Once you have published your model, its symbol will switch from to in the Published column of the Administration Page.
The Sharing column indicates whether you have shared your model publicly or privately .
Generally, you can publish your Bayesian network model as a public or a private Questionnaire. This is analogous to the options for publishing a WebSimulator.
The dates show the respective expiration dates of your subscriptions.
A public Questionnaire will also appear as a tile on the public WebSimulator homepage at . This is a great way to showcase your work to the wider world.
A pop-up window shows all settings specified in the and also features the same tabs:
Clicking the Edit Properties button allows you to modify the properties that you previously specified in the or, before that, in the .
This is now the live version of the Questionnaire, of which you could previously only launch a static preview from the .
Open a web browser and go to
On the WebSimulator homepage at , click the navigation icon , then select Go To Adaptive Questionnaire .
Clicking on the information icon brings up the Description of the current Adaptive Questionnaire.
Clicking on the eye icon displays the Image associated with the current Adaptive Questionnaire.
If you cannot answer the most relevant question, as suggested by the Input position, you can skip this question by clicking the icon .
Clicking the eraser icon removes all pieces of evidence that you have set on the Inputs.
To highlight the impact of your observations on the Output, you have to select the option Display Variations available by clicking the Options icon in the upper right corner of the page.
For gauge-style Outputs, the variations are also highlighted with green and red arcs for positive and negative impacts, respectively.
Click the gear icon to open the Options drop-down menu.
Select Store Reference State
.
To save an Observation Set, click the plus icon from the Icon Bar.
Now, an additional icon in the Icon Bar indicates the presence of your saved Observation Set.
Click the Plus icon again, and you can add to the same Observation Set Name or create a new Observation Set Name.
If you click on the Observation Set icon in the Icon Bar, a drop-down menu lists all saved Observation Sets with their name and sequence number.
Click the gear icon, then select Observations > Clear Stored Observations
.
Click on the gear icon in the Icon Bar, then select Observations > Export Stored Observations
.
Click on the gear icon in the Icon Bar, then select Observations > Import Observations
.
Click on the gear icon in the Icon Bar, then select Simulation Mode .
With the Simulation Mode activated, the Icon Bar features two additional icons as scroll buttons: .
Click the gear icon, then select Layout > Input or Layout > Output
and pick the desired layout for each panel:
Open a web browser and go to the WebSimulator homepage at
On the WebSimulator homepage at , click the navigation icon in the Icon Bar, then select Go to Simulator .
Clicking on the information icon brings up the Description of the current WebSimulator.
Clicking on the eye icon displays the Image associated with the current WebSimulator.
Moving a slider
Selecting an item from a drop-down menu
Selecting a switch
Typing in a numeric value
You click the Eraser icon in the upper right corner of the page. This erases all pieces of evidence set thus far.
To highlight the impact of your observations on the Outputs, you have to select the option Display Variations
available by clicking the Options icon in the upper right corner of the page.
For gauge-style Outputs, the variations are also highlighted with green and red arcs for positive and negative impacts, respectively.
Click the Gear icon to open the Options drop-down menu.
The deltas shown on the Outputs now are calculated based on the Reference State .
To save an Observation Set, click the Plus icon from the Icon Bar.
Now, an additional icon in the Icon Bar indicates the presence of your saved Observation Set.
Simply click the Plus icon again, and you can add to the same Observation Set Name or create a new Observation Set Name.
If you click on the Observation Set icon in the Icon Bar, a drop-down menu lists all saved Observation Sets with their name and sequence number.
Click the gear icon, then select Observations > Clear Stored Observations
.
Click on the gear icon in the Icon Bar, then select Observations > Export Stored Observations
.
The format of this Scenario File is identical to the produced by BayesiaLab. As such, it is an ideal format for exchanging scenarios for review and discussion between BayesiaLab users and WebSimulator users.
Click on the Gear icon in the Icon Bar, then select Observations > Import Observations
.
Click on the gear icon in the Icon Bar, then select Simulation Mode .
With the Simulation Mode activated, the Icon Bar features two additional icons as scroll buttons: .
Click the gear icon, then select Layout > Input
or Layout > Output
and pick the desired layout: