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.

Release of BayesiaLab 4.6

June, 15 2009 / software release

The main new features of this release are:

• 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. The resulting probability distribution can be fixed or not;
• Target Dynamic Profile (Analysis - Report - Target Analysis menu): this tool exploits now the new mean value evidence feature to compute more realistic action policies. The levers can now be evaluated not only with hard evidences, but also with soft evidences corresponding to smooth evolution of the levers levels (e.g. 10% of levers' improvement). This tool allows now saving the policies' actions in scenario files;
• Target Mean Analysis (Analysis - Graphic menu): this new tool is a graphical extension of the Total Effect analysis. It allows to graphically describe the relation (linear and non linear) between the variable values and the target values over the entire variation domain of the variable;
• Influence Analysis with respect to the Target Node (Analysis - Graphic menu): generalization of the Influence Analysis wrt the Target state tool that was available thanks to the node contextual menu. This new tool works on a set of nodes, for all the states, and can be used in both directions, i.e. to analyze the P(Target|node) or P(node|Target);
• Quadrants: this new 2-dimensional graph depicts the node significance on the Y-axis versus its mean value on the X-axis. The quadrants are defined by using the mean of the significance values and the mean of the variable values. In Satisfaction Analysis, the upper left quadrant corresponds to the levers that have to be improved (high importance and low mean value), the upper right quadrant corresponds to the levers that have to be maintained (high importance and high mean value), the lower right quadrant corresponds to the levers that can be overkilled or that corresponds to "must have" characteristics, and the lower left quadrant contains the levers with low priority (low importance and low mean value). This graph can be generated by using the Target Report analysis (Analysis - Report - Target Analysis - Correlations with the Target node) or by using the Total Effect analysis (Analysis - Report - Target Analysis - Total Effects on Target). In the first case the node significance corresponds to the Relative significance of the node computed with the Mutual information. In the second case the node significance is based on the Standardized Total Effect or the Total Effect;
• Discretization: The discretization toolkit is now available in the node editor. Discretization can then be modified after the database loading, manually by using the distribution function or the density function, or automatically by using the Decision tree, the KMeans, the equal frequency and the equal distance algorithm;
• Zoom on the monitors (Toolbar icons): it is now possible to change the size of the monitors. This is particularly useful during the video-presentations: zoom-in to increase the readability, zoom-out to allow a global dashboard;
• Parallelization of learning algorithms, multiple clustering, etc., in order to benefit from the computers with several cores or processors;
• and, as usual, increased ergonomics of the interface and performance improvements.

Click here to consult the newsletter that describes more in details all the new functionalities of BayesiaLab 4.6 »

Download from now on BayesiaLab 4.6 for a new 30 days evaluation period »