www.bayesia.com does not fully support your browser (Internet Explorer 6).
We suggest upgrading to IE 7 or downloading Firefox for a more enjoyable web experience.

BayesiaLab 4.6: New Features

Data Evidence scenario file saved with graph Junction tree saved with graph Fixed probabilities saved with graph Arc dictionaries Node position dictionaries Fixed arc dictionaries Arc comment dictionaries Arc color dictionaries Syntax of dictionaries modified Merging filtered values with not filtered state * at association Network State virtual number Local structural coefficient Exclusion as a node property Fixed arc as an arc property Select disconnected nodes Learning Reports automatically saved in multiple clustering Weighting cluster state values by each node's binary mutual information Factor node included in class Factor in multiple clustering Factor's filtered state automatically detected in clustering and multiple clustering Contingency table fit and deviance in multiple clustering's final report Sub-network's factor not observable in multiple clustering Summary of results in multiple clustering's report Clustering's and multiple clustering's settings integrated into wizards Parallelization of the multiple clustering Parallelization of structural learning algorithms Improvement of learning algorithms Inference Evidence scenario file for batch likelihood Indicator when fixed probabilities did not converge Manual selection of an evidence set in a scenario file Analysis Target dynamic profile with directed variation of the mean Compute prior variations in target dynamic profile Option to associate the generated scenario file in dynamic profile Filtered states not used as driver in dynamic profile Button to save scenario in dynamic profile Influence analysis wrt target node Target mean analysis Filtered states excluded from target state optimization Target state optimization with scenario files for saving and association Quadrants chart in correlations with target node report R and R2 indices computed in targeted network performance Contingency table fit and deviance for global network performance Contingency table fit and deviance for correlations with target node report Node means added in correlations with target node report Buttons to validate arc's comments and colors in arc force and Pearson analysis Quadrants chart for total effects report Computation of total effects modified Monitors New evidence setting with mean value Zoom on monitors Monitor finder Fixed probabilities not set if no convergence Mean, standard deviation and value in the monitor copy Interface Discretization of continuous nodes in the node editor Automatic state renaming when modifying intervals Values automatically computed when modifying intervals Button to generate values in the node editor Indicators of filtered state and missing values on nodes New icons for validity and comments Menu and button to hide all indicators Contextual menu to remove the selected comments Toolbar for network loading and saving Selection kept after closing search dialog Main window's extended state saved All charts become resizable Tools Global precision computation modified in cross-validation Possibility to save networks in comparison tool Option for saving predicted/existing values in the targeted cross-validation Shortcuts Tooltips on node's indicators and comment indicators by pressing I Tooltips for network's comment by pressing W Tooltips on monitors by pressing I CTRL + A to select all monitors
Data
Evidence scenario file saved with graph	The evidence scenario file associated with the network can now be saved inside the same file as the network. This operation is now realized by default in BayesiaLab; nevertheless, an option allows deactivating automatic evidence scenario file saving in the "Settings" menu. When a loaded network contains an evidence scenario file, this one is automatically loaded too (unless the user explicitly specifies not to do so).

Junction tree saved with graph	If a junction tree has been built for a graph, it can be saved in the same file as the graph. When the graph is loaded, the junction tree is automatically re-created from the file. This way is much quicker than creating the junction tree from scratch, in particular when the graphs are big and heavily connected.

Fixed probabilities saved with graph	If some probability distributions are observed as fixed, they now can be saved when the network is saved, like hard and soft evidences. However, you must take care if they cannot converge the first time the inference is done.

Arc dictionaries	Arc dictionaries were added to BayesiaLab. It is possible to export all the arcs in a file and to import them as well. Importing arc dictionaries allows associating a set of arcs to the network. The indicated arcs can be added or removed from the network. The arc removal will always be done before adding an arc. Before adding an arc, all the constraints belonging to the Bayesian network as well as the arc constraints and the temporal indices will be checked. If a constraint is not verified, then the arc won't be added. The syntax is: Name of the arc's starting node or class, ->, <- or even -- to indicate the both possible orientations, name of the arc's ending node or class, Equal, Space or Tab, true for an added arc or false for a removed arc. The last occurrence is always chosen.

Node position dictionaries	The different nodes' positions can be saved in a dictionary or loaded from it. The syntax is: Name of the node Equal, Space or Tab, position. The position is represented by two real numbers separated by a Space. The first number represent the x-coordinate of the node and the second number the y-coordinate. A node can be present only once otherwise the last occurrence is chosen.

Fixed arc dictionaries	The fixed arcs defined in the network can be exported to a dictionary. Loading a dictionary allows you to define if some arcs are fixed or not. The syntax is: Name of the arc's starting node or class, ->, <- or even -- to indicate the both possible orientations, name of the arc's ending node or class, Equal, Space or Tab, true for an fixed arc or false for a not fixed arc. The last occurrence is always chosen.

Arc comment dictionaries	Arc comments can now be saved in a dictionary or loaded from it. The syntax is: Name of the arc's starting node or class, ->, <- or even -- to indicate the both possible orientations, name of the arc's ending node or class, Equal, Space or Tab, comment. The comment can be any character string without return (in html or not). The last occurrence is always chosen.

Arc color dictionaries	The arc colors can be exported in a dictionary or imported from it. The syntax is: Name of the arc's starting node or class, ->, <- or even -- to indicate the both possible orientations, name of the arc's ending node or class, Equal, Space or Tab, color. The color is defined as Red Green Blue 8 bits by channel color written in hexadecimal (web format). For example green gives 00FF00, yellow gives FFFF00, blue gives 0000FF, pink gives FFC0FF, etc. The last occurrence is always chosen.

Syntax of dictionaries modified	As the dictionaries are now saved in UTF-8, all special characters can be used. However, if the key part of a line, i.e. the part before the separator character (Equal, Space or Tab), contains any of these separator characters, those characters must be written with a \ (backslash) character before in the text file: for example the node named Visit Asia will be written Visit\ Asia in the file if it is used as key. On the right part of the line, after the separator character, the only forbidden character is the new line as it defines the end at a line in the file.

*Merging filtered values with not filtered state at association**	Sometimes filtered values exist in the database but are not declared in the network. Now, it is possible to merge them with the specific state *, if it exists. In this case, this state will be automatically defined as filtered for each concerned node.
Network
State virtual number	The state virtual number is a new property added to nodes. It allows replacing the real number of states during the learning with the MDL score. The node's state number has an important impact on the MDL score computed during the structural learning. This allows influencing the network's structural complexity locally to the node. More a node has states less it has chance of having linked parents during the learning and vice versa. Decreasing this parameter decreases the MDL score of the node and vice versa. This property can be changed with the node editor: The Propagate state virtual number button displays a dialog box allowing selecting the classes, associated to the node, on which we want to propagate the state virtual number. In this case, all the nodes belonging to these chosen classes will have the same state virtual number or any if there is no one. If one node has a state virtual number defined, the icon is displayed in the network's status bar. A click on this icon displays the global state virtual number editor, regrouping each node with its corresponding state virtual number: This editor can be displayed as well with the graph's contextual menu.

Local structural coefficient	The local structural coefficient is a new property added to nodes. This parameter acts like the network's global structural coefficient but is proper to each node. It can increase or decrease the structural complexity of the network at the node. This parameter acts on the whole MDL score of the node contrary to the state virtual number. More a node has a high MDL score the less it has chance of having linked parents during learning and vice versa. Decreasing this parameter decreases the node's MDL score and vice versa. This property can be changed with the node editor: The Propagate local structural coefficient button displays a dialog box allowing selecting the classes, associated to the node, on which we want to propagate the local structural coefficient. In this case, all the nodes belonging to these chosen classes will have the same local structural coefficient or any if there is no one. If one node has a local structural coefficient different from 1, the icon is displayed in the network's status bar. A click on this icon displays the global node's structural coefficient editor, regrouping each node with its corresponding coefficient: This editor can be displayed as well with the graph's contextual menu.

Exclusion as a node property	The exclusion is now integrated as a node property. It can be changed in the node editor: The Propagate exclusion button displays a dialog box allowing selecting the classes, associated to the node, on which we want to propagate the exclusion status. In this case, all the nodes belonging to these chosen classes will be excluded or not. The property can be modified in the section Properties of the node's contextual menu. If several nodes are selected, the exclusion status is propagated to these nodes. If the current node belongs to one or more classes, a dialog proposes also to propagate the exclusion to the classes, i.e. to the nodes belonging to these classes.

Fixed arc as an arc property	Now, fixing an arc is done with the corresponding item in the section Properties of the arc's contextual menu. If several arcs are selected, the property is propagated to these arcs.

Select disconnected nodes	In the Edit>Select menu, it is possible to select all the disconnected nodes by clicking on the Disconnected item.
Learning
Reports automatically saved in multiple clustering	In the multiple clustering, the intermediate reports are automatically saved in the output directory.

Weighting cluster state values by each node's binary mutual information	When numerical clusters are computed in clustering, the numerical values are multiplied by the binary mutual information between this state and each node. By this way, the impact of each node on each cluster state is better represented than before.

Factor node included in class Factor in multiple clustering	Now, each created factor node in multiple clustering is included in its corresponding class.

Factor's filtered state automatically detected in clustering and multiple clustering	When a factor node is created in clustering or multiple clustering, it is possible that one of the cluster states represents the filtered states of the nodes. Now, a new algorithm tries to find which cluster represents the filtered states. If a state is found, it is declared as filtered in the factor node.

Contingency table fit and deviance in multiple clustering's final report	Two new performance indices are computed in the multiple clustering's final report. They are the same as in the Correlations' with target node report. Contingency Table Fit: Used to represent the degree of fit between the network's joint probability distribution and the associated data. More the network correctly represents the database, the more the value tends towards 100%. This measure, computed from the database's mean log-likelihood is equal to 100% when the joint is completely represented in like in the fully connected network or to 0% when the joint is represented by a fully disconnected network. The dimensions represented by the not observable nodes are excluded from the computation. Deviance: This measure is computed from the difference between the network's mean log-likelihood and the database's mean log-likelihood. More the value tends towards 0 the more the network is close to the database.

Sub-network's factor not observable in multiple clustering	In each network created for each class in multiple clustering, the factor node is declared as not observable.

Summary of results in multiple clustering's report	At the beginning of the report, a summary indicates the number of factors found, and the minimum, average and maximum number of clusters, mean purity and contingency table fit.

Clustering's and multiple clustering's settings integrated into wizards	The following clustering parameters are now directly available in the clustering and multiple clustering wizards: Maximum Drift: indicates the maximum difference between the clusters probabilities during learning and those obtained after missing value completion, i.e. between the theoretical distribution during learning and the effective distribution after imputation over the learning data set. Minimum Cluster Purity in Percentage: defines the minimum allowed purity for a cluster to be kept. Minimum Cluster Size in Percentage: defines the minimum allowed size for a cluster to be kept.

Parallelization of the multiple clustering	The multiple clustering algorithm is now parallelized in order to benefit from the computers with several cores or processors.

Parallelization of structural learning algorithms	Each structural learning algorithm has been parallelized in order to benefit from the computers with several cores or processors. These algorithms include Taboo, EQ, SopLEQ, Maximum Spanning Tree and Taboo Order. The performance has been dramatically improved. If the computer has only a single core single processor, the speed is not negatively impacted.

Improvement of learning algorithms	Different optimizations increase the speed of the learning algorithms, whatever the computer is. For example, in certain cases, combined with parallelization, the running time of Taboo has been divided by 10, EQ by 3, etc.
Inference
Evidence scenario file for batch likelihood	Now, the batch likelihood can be performed on the associated evidence scenario file like the other batch analysis.

Indicator when fixed probabilities did not converge	As the probability fixation may sometimes not converge, i.e. it is impossible to obtain the given probability distribution, a message is written in the console and a warning dialog box is displayed. This can append by setting manually the distribution or with evidence scenario file in temporal, interactive inference and interactive updating.

Manual selection of an evidence set in a scenario file	When an evidence scenario file is associated to the network, a list allows choosing one of the evidence sets contained in the file. This list can be displayed in four different ways: Interactive inference: a right-click on the step text field displays the list. Once the wanted evidence set is selected, the corresponding evidences are set. Interactive updating: a right click on the index text field displays a list of the evidence sets contained in the file. A click on a line performs the updating starting from the current index up to the specified index, taking into account the corresponding evidences. Temporal: a right click on the index text field displays a list of the evidence sets contained in the file. A click on a line performs the temporal simulation from the current index to the specified index, taking into account the corresponding evidences. Manual observation: a right click on the icon located in the network's status bar displays a list of the evidence sets contained in the file. Once the wanted evidence set is selected, the corresponding evidences are set. This is available in validation mode only.
Analysis
Target dynamic profile with directed variation of the mean	New search methods were added to the target dynamic profile report. Three search methods for criterion optimization are now available: Hard Evidences: Only hard evidences will be used for the optimization. Value/Mean Variations in %: On each driver variable a probability distribution will be applied in order to vary its mean by plus or minus the given percentage. This percentage can vary between 0 and 1000%. The computed mean is bounded by the variable's variation domain's limits. Each node's mean is computed from the values associated to the states. If there is no associated value, if the node is continuous, its mean will be computed from the intervals, and if the node is discrete with integer or real states, the mean is computed from them. If there is no possibility to compute the mean, a default set of values from 0 to the number of states minus one is used. Value/Mean Variations in Domains %: On each driver variable a probability distribution will be applied in order to vary its mean by plus or minus the given variable's domain's percentage. This percentage can vary between 0 and 100%. The computed mean is bounded by the variable's variation domain's limits. Each node's mean is computed from the values associated to the states. If there is no associated value, if the node is continuous, its mean will be computed from the intervals, and if the node is discrete with integer or real states, the mean is computed from them. If there is no possibility to compute the mean, a default set of values from 0 to the number of states minus one is used.

Compute prior variations in target dynamic profile	A new option allows computing only prior variations, i.e. without cumulative effect: each previous observation done will be removed before finding the best following evidence.

Option to associate the generated scenario file in dynamic profile	A new option allows associating to the network the evidence scenario file corresponding to the found results. If an evidence scenario file already exists, it can be replaced by the new one or the evidences can be appended to it.

Filtered states not used as driver in dynamic profile	Whatever the used search method in target dynamic profile, the filtered states are excluded from the search.

Button to save scenario in dynamic profile	In the report's window, in addition to the usual saving and printing options, a button Save Scenario allows you to save in a file the evidence scenario corresponding to the chosen optimization.

Influence analysis wrt target node	This tool allows visualizing, for each target node's state, the influence of the other nodes in the network. This analysis is done only on the selected nodes or on all the nodes if no nodes are selected. If some nodes are not connected to the target, directly or not, they will be excluded from the analysis. This graph is dynamic, i.e. evidences can be set on the monitors and the graph will be updated in order to reflect the modifications. The evidence context is displayed at the top of the graph, under the title. There are two kinds of display that can be chosen up to the checkbox Inverse Influence: Display of each target's state's probability knowing each state of each node: The target's state's prior probability is shown by a red line. Display of each state's probability for each node knowing each one of the target's state: For each node, each state's prior probability is shown by a red line. In this mode, the nodes' current values and delta are displayed in each "monitor". This value is computed from each node's values if they exist. If not, they are computed from the data for continuous nodes or from the intervals if no data is associated. If the node has real or integer states, they will be used. Otherwise, integer values are automatically generated, starting from 0. A checkbox Compute from Database is displayed if a database is associated to the network. It performs the analysis not by using the Bayesian inference anymore but by using the probabilities directly computed from the database. When the mouse is moved over a bar chart, a tooltip displays the state's name and the associated exact probability. It is possible to save the graph's image in a file or to print it. The graph's contextual menu allows the user to: Sort the displayed nodes within three modes: Default Order: the nodes' initial order is used or the displayed monitors' order, if they are monitored Sort by Global mutual Information Sort by Binary Mutual Information Display Comment Instead of Name Display States' Long Name Copy the image of the graph

Target mean analysis	This tool allows you to graphically view the impact of changes in the selected nodes' means on the target node's mean. This lets you see the relationship between each node and the target variable in the form of curves. For each node, its mean will vary from the minimum to the maximum of the variation domain and determine, for each variation, the corresponding probability distribution up to MinXEnt method, in the same manner as for the observation of a node's mean in the monitors. Each node will be observed with this probability distribution and the corresponding value of the target node is calculated. For each node the mean is computed from the values associated with the node: if the node has values associated with its states, the mean is computed from them, otherwise if the node is continuous, its mean is computed from the intervals, and if the node is discrete with integer or real states, the mean is computed from them. If there is no possibility to compute the mean, a default set of values from 0 to the number of states minus one is used. A dialog box lets you configure the display of curves: For the target variable, it is possible to display either the actual target's mean or the delta of the mean with the prior target's mean. Likewise for other variables, it is possible to display either the actual mean or the delta of the mean with the prior mean. However, it is also possible to display the real variables' means, but all standardized between 0 and 100. The option Use Hard Evidences replace the previously used algorithm to compute the points of each curve simply by observing each node states and by determining the node's mean and the corresponding target's mean. The option Order by Strength sorts the variables from the slope the more towards 90 degrees to the slope the more towards -90 degrees. This sort variables that have the most positive impact on the target's mean to the most negative impact. This slope is calculated at the midpoint of the curve. By default, the nodes are sorted according to the order of the corresponding monitors, if they exist. This figure represents the result of the previous settings. The legends are sorted according to the slope of the curves. When the mouse hovers on any of the points on the curves, a tooltip displays the name of the variable represented by the curve. The coordinates are displayed at the top of the window. The evidence context, if it exists, is also displayed at the top of the window. The following figure represents the display of the deltas. A zoom has been done on the central points. The zoom is operated as in the scatter graph. It is also possible to change the curve's color by clicking on the colored square to the left of the node names in the legend. The chart's contextual menu allows the user to: Display Comment instead of Name Copy the image of the chart or the curves' points as text or html Print the image of the chart

Filtered states excluded from target state optimization	In the target state optimization, the existing filtered states are excluded from the search.

Target state optimization with scenario files for saving and association	In the target state optimization, it is now possible to associate the results as evidence scenario file: you simply have to choose the number of examples you want to associate. If you stop the search before attaining the wanted number of examples, the associated evidence scenario file will be smaller. You also can save the found tuples in an evidence scenario file, on the disk, by specifying its name.

Quadrants chart in correlations with target node report	A new button named Quadrants has been added to the report of the correlations with target node. This button displays the Quadrant chart of the node's relative significance with the target relatively to each node's mean: The points represent the variables. If a node has a color then the point will be displayed in this color. The node's name is displayed at the right of the point. When moving the mouse on the point, the coordinates are displayed in the top panel. The top panel shows the number of points displayed as well as the evidence context. It is possible to zoom in the chart as in the scatter of points. The chart's contextual menu allows the user to : Display or not the nodes' comment instead of their names Copy the chart as image or as table of points (text or html) Print the chart The chart is automatically resized when the size of the window changes. This chart is cut into four quadrants whose separations are: Along X-axis: the mean of the variables' means Along Y-axis: the mean of the relative significance of the nodes with the target Quadrant 1: Top right: contains the important variables greater than the mean Quadrant 2: Bottom right: contains the important variables few or not important but greater than the mean Quadrant 3: Bottom left: contains the few or not important variables that are below the mean Quadrant 4: Top left: contains the important variables but that stay below the mean It is possible to save the image of the chart in a file with the corresponding button.

R and R2 indices computed in targeted network performance	Two new indices are displayed in the targeted network performance window. The Pearson coefficient and its square value are displayed at the top of the window. They are used to measure the quality of the numerical prediction:

Contingency table fit and deviance for global network performance	Two performance indices are added to the global network performance charts: Contingency Table Fit: Used to represent the degree of fit between the network's joint probability distribution and the associated data. More the network correctly represents the database, the more the value tends towards 100%. This measure, computed from the database's mean log-likelihood is equal to 100% when the joint is completely represented in like in the fully connected network or to 0% when the joint is represented by a fully disconnected network. The dimensions represented by the not observable nodes are excluded from the computation. Deviance: This measure is computed from the difference between the network's mean log-likelihood and the database's mean log-likelihood. More the value tends towards 0 the more the network is close to the database.

Contingency table fit and deviance for correlations with target node report	When a database is associated to the network, two performance indices are displayed in the correlations with target node report: See previous item for explanations.

Node means added in correlations with target node report	The node means are computed, in the correlations with target node report, for each node according to the target node and according to each target node's state. Each node's mean is computed like this: if the node has values associated with its states, the mean is computed from them, otherwise if the node is continuous, its mean is computed from the intervals, and if the node is discrete with integer or real states, the mean is computed from them. If there is no possibility to compute the mean, a default set of values from 0 to the number of states minus one is used.

Buttons to validate arc's comments and colors in arc force and Pearson analysis	A new button was added to the arc force analysis' and Pearson analysis' toolbars . It stores the current comment and color associated to the arc by the analysis. Comment and color replace existing ones. Arc force analysis: Pearson's correlation analysis:

Quadrants chart for total effects report	In addition to the saving and printing classical options, a button Quadrants allows displaying a Quadrant chart of the target's mean relatively to the total effect or the standardized total effect. The choice is done in the following dialog box: The obtained result is purely illustrative, it can't really be analyzed: The points represent the variables. If a node has a color then the point will be displayed in this color. The name of the node is displayed at the right of the point. When moving the mouse on the point, the coordinates are displayed in the top panel. The top panel shows the number of points displayed as well as the evidence context. The chart's contextual menu allows the user to : Display or not the nodes' comment instead of their names Copy the chart as image or as table of points (text or html) Print the chart The chart is automatically resized when the size of the window changes. This chart is cut into four quadrants whose separations are: Along X-axis: the mean of the variables' means Along Y-axis: the mean of the criterion's importance (here the standardized total effect on the target) Each quadrant has a specific meaning according to the knowledge represented by the Bayesian network (marketing, satisfaction, etc.). In a general way, the quadrants are: Quadrant 1: Top right: contains the important variables greater than the mean Quadrant 2: Bottom right: contains the important variables few or not important but greater than the mean Quadrant 3 : Bottom left: contains the few or not important variables that are below the mean Quadrant 4 : Top left: contains the important variables but that stay below the mean It is possible to save the image of the chart in a file with the corresponding button.

Computation of total effects modified	The MinXEnt algorithm used to compute the total effects was modified in order to be more accurate and increase correctness.
Monitors
New evidence setting with mean value	This new feature proposes a new way for entering evidence in a Bayesian network. It allows computing a new probability distribution based on a numerical target mean value. The obtained probability distribution minimizes the difference with respect to the initial probability distribution. This can then be interpreted as the easiest action to do to get this expected value. When a node has values associates to states or is continuous, it is possible to choose a target mean/value for this node. An algorithm based on MinXEnt allows determining a probability distribution corresponding to this target mean/value, if this distribution exists. Of course, the indicated target value must be superior or equal to the minimum value and superior or equal to the maximum value. The checkbox Fix Probabilities allows indicating if the distribution found must be set as fixed probability distribution or as likelihoods, in the same way as for the probability setting.

Zoom on monitors	In the monitor toolbar, three buttons have been added to perform zoom on monitors. This feature is particularly useful during the presentation with video-projectors (zoom in) or to display all the monitors at the same time (zoom out). zoom in monitors zoom out monitors default zoom

Monitor finder	A new tool used to find monitor has been added to the node and arc finder. The thumbnail Monitors displays the monitor search user interface. Obviously, you must be in validation mode and having monitored nodes. The editable combo box allows entering the name of the node or of its class that we are searching. We can use also the special characters * and ?: The character * represents a series of 0 to n unspecified characters. The character ? represents exactly one unspecified character. The check box "Case sensitive" means that the capital and small letters will be interpreted as is. The search options allow searching among the nodes only, the classes only or both of them. Examples: if we want to find all the monitors of nodes beginning by "Cap", we will enter the following search: "Cap" (without the quotes) if we want to find all the monitors of nodes containing the letters "A" and "B" separated by a single character, we will enter the following search : "A?B*" (without the quotes) Once the search is done, the list of the found monitors and the number of monitors are displayed. By selecting a monitor in the list, this one begins to blink in the monitor windows. If the window is closed, the current selection remains displayed.

Fixed probabilities not set if no convergence	When a fixed probability distribution is set manually, to obtain the indicated distribution, a convergence algorithm is used. However, sometimes this algorithm cannot converge towards the target distribution. In this case, the probabilities fixing is not done and the node comes back to its initial state. In this case a warning dialog box is displayed and an information message is also written in the console.

Mean, standard deviation and value in the monitor copy	In addition to the probabilities, the mean, standard-deviation and value are also copied if they are present, in plain text or html format.
Interface
Discretization of continuous nodes in the node editor	In the node editor, a new tool is available for continuous nodes. By clicking on the "Curve" at the top of the panel, the discretization from data interface is displayed. This button is displayed only if data are associated to this node: This interface is similar to the manual discretization interface of the data import/association. It represents the current node's data distribution function. The X-axis represents the number of individuals and the Y-axis represents the values of the continuous variable. The user can switch the view of the data to a representation of the density curve generated by the Batch-Means method. In this view, the data's density curve is displayed. The continuous variable's values are represented along X-axis and the density of probability is represented along the Y-axis. The two red areas at each extremity indicate that the curve may not be accurate and can't be used to place here some discretization points. This window is fully interactive and allows, in both view: Adding a threshold: Right Click Removing a threshold: Right Click on the threshold Selecting a threshold: Left Click on a threshold Moving a threshold: Left Click down and mouse move, the current Y-Coordinate appears in the Point box : Zooming: Ctrl + Left Click down + move + release to define the area that we want to enlarge. In the distribution function, the zoom will be done vertically and in the density curve, it will be done horizontally. It is possible to zoom successively as much as you need. Unzooming: Ctrl + Double Left Click When a threshold is manually moved, the superior and inferior intervals' sizes, taking into account the weights, are displayed on both sides of the line. Besides this distribution function, the button: allows having access to the three automatic discretization methods through a new dialog. This part can be considered a wizard for the manual discretization as it is possible to launch these methods, to see the resulting discretization on the distribution function, and then to modify the result by moving, deleting and adding thresholds. If the chosen discretization fails, a dialog box is displayed to warn the user. In this dialog it can change the chosen discretization. When an interval is modified, deleted or added, whatever the way to do this, if data are associated to the node, then the modified interval values are automatically updated from the data. In the same way, the modified interval names are automatically generated if they are numerical. If the new manual or automatic discretization is validated by clicking the button Accept of the editor, the database will be automatically updated in order to take into account this new discretization. It will be also stored as a manual discretization in the database.

Automatic state renaming when modifying intervals	When an interval is modified, deleted or added, whatever the way to do this, then the states that are modified had their names automatically updated to represent the interval, but only if theses names are numerical.

Values automatically computed when modifying intervals	When an interval is modified, deleted or added, whatever the way to do this, if data are associated to the node, then the modified interval values are automatically updated from the data.

Button to generate values in the node editor	It is now possible to automatically generate values associated to each node's state by clicking on the Generate Values button. This button is visible only if the node is continuous and has continuous data associated to it. Each value is the mean of the values in each interval. Weights are taken into account.

Indicators of filtered state and missing values on nodes	In addition to the warning and error indicators already present on a node, two new icons have been added: Filtered state indicator: indicates that one state of the node is declared as filtered. Missing values indicator: indicates that there are some missing values in the data associated to this node. When the mouse hovering this indicator, while pressing I, a tooltip displays the number of missing values associated to this node.

New icons for validity and comments	The icons for the warning, error and comment indicators have been redone : Warning: Error: Comment: When the mouse hovering these indicators, while pressing I, a tooltip displays the message associated to the indicator: warning message, error message and comment message. You can do the same on the arc's comment indicator.

Menu and button to hide all the indicators	A new button in the view toolbar and a new menu hide information in the view menu allow hiding all the displayed indicators. It allows the graph to be displayed clearly. The warning, error, node's comment and arc's comment indicators are hidden. They can be displayed again by clicking on this same menu/button.

Contextual menu to remove the selected comments	In the node's and arc's contextual menus, in the properties section, a new item allows the user to remove the comment of all selected nodes or arcs at the same time. For the nodes, a dialog box proposes to propagate the removal to the classes the node belongs.

Toolbar for network loading and saving	A toolbar has been added in the network file chooser. As the network's file can contain a n associated database, an associated evidence scenario file and the network's junction tree, three icons indicates if they are present or not in the file. For loading, if a database, a scenario file or a junction tree are present in the network's file, the corresponding icons are enabled. You check them if you want to load the database, scenario file or junction tree, or uncheck them if not. for loading associated database for loading associated evidence scenario file for loading associated junction tree For saving, if you it is the first time the network is saved or if the network is saved as..., you can choose if you want to save the associated database, scenario file or junction tree if they exist. The icons are the same and enabled if the saving is possible. You can check or uncheck them. for saving associated database for saving associated evidence scenario file for saving associated junction tree

Selection kept after closing search dialog	Now, when the finder dialog is closed, the last selections done are kept for nodes, arcs and monitors, so you can close the finder without losing the selections.

Main window's extended state saved	Now, the main window's extended state is saved when the BayesiaLab's main frame is closed. When the window is in extended state and is closed, when it is reopened, the window is now in its extended state and its normal size is not overridden. Then if you set the window back to its normal state, the previous normal state is used instead of the full screen size. For example, your window's size is 640 * 400 in normal state. Then you click on the extended state (full screen) then the window goes to extended state and its size is now the size of the screen (1650 * 1080 for example). After that, BayesiaLab is closed. Once re-launched, the BayesiaLab's window goes back to extended state and the window occupies 1650 1080. If you click on the button to go back to the normal state, the window's size will be 640 480.

All charts become resizable	All charts are now completely resizable in BayesiaLab. So, if you extend the containing window, the chart will be computed again to fit the window's size.
Tools
Global precision computation modified in cross-validation	In cross-validation, the computation of the global precision has been modified. Instead of computing the mean of each sample's precision, the global precision is now computed with each case of the database. This value is more correct than the other, in particular when the number of cases in each sample was small.

Possibility to save networks in comparison tool	A new toolbar has been added in the network comparison tool. This tool is used in the network comparison, targeted cross-validation and arc confidence cross-validation. A new icon allows the user to save the current displayed network. This option is only available for comparison structures. It is disabled for Synthesis and Reference structures.

Option for saving predicted/existing values in the targeted cross-validation	In the targeted cross-validation's report, in addition the previous Save, Print, Graphs and Extract Network options, a new Save Values button allows saving into a file the numerical values of the target predicted by each network on its corresponding test set.
Shortcuts
Tooltips on node's indicators and comment indicators by pressing I	Now, to obtain the messages corresponding the warning, error, missing values and comment indicators, you must press while hovering the indicator's icon with the mouse. This can be done on node's and arc's indicators.

Tooltips for network's comment by pressing W	To obtain the tooltip for the network's comment, you must press while the mouse hovers on the network's background.

Tooltips on monitors by pressing I	As for the node's or arc's indicators, you must press while hovering the monitor to display the tooltip containing the exact values for probabilities, mean, standard-deviation, etc.

CTRL + A to select all monitors	When the focus is in the monitor area (by clicking on a monitor, for example), you can select all the monitors with the shortcut: + . On Mac OS X, the Ctrl key is replaced by the Command key.

BayesiaLab 4.6: New Features

Data

Network

Learning

Inference

Analysis

Monitors

Interface

Tools

Shortcuts