       |
|
News Letter: November 2004| Bayesia Market Simulator | Bayesia Engines | Bayesia Graph Layouting Engine |
BayesiaLab: the
same framework for your Decision Support and Data Mining tasks
Main new features available in BayesiaLab 3.2
- A priori knwoledge expressed with an initial Bayesian network
is taken into account by all the learning algorithms
- New analysis report that returns a numerical measure
of the importance of the probabilistic relations (arcs). This report
then complete the visual analysis based on the thikness of the arcs
- New missing values imputation tool that use all the available
information to complete missing data in a rigourous fashion
- Confidence measure of the automatically estimated probabilities
by displaying the number of examples used for the estimation
of each probability
- New symmetric algorithm for a very efficient layouting
of the networks that are moderatly connected
- New search tool that allows searching nodes
and arcs that can be described with meta characters
- Arcs can have associated color tags and comments
- Improvement of the importation wizard for data coming from
Data Bases
- New tool for the generation of data corresponding to a Bayesian network that includes the possibility to specify a percentage of missing values
Bayesia Market Simulator:
The necessary tool for efficient market simulations
Main characteristics of Bayesia Market Simulator 1.1
- Learning of a Bayesian network with BayesiaLab
to model the choice of the individuals with respect to their
characteristics and those of the products
- Description of the market by using a file of individuals
or by generating a market based on the Bayesian network
that models the choice
- Scenarios specification by describing the offers
and the market segments
- Similarity rules to indicate that some products are close,
and that the rejection of an offer belonging to a set of similar product is
valid for all the products of that set
- Efficient computation of the market shares corresponding
to each product as well as the probability associated to the rejection
of all the offers
- Saving, for each individual, of the probabilties associated
to the products and the rejection
- Generation of log files (text and HTML) to precisely record
the simulations that have been carried out
Bayesia Engines: Bayesian networks
directly available in your applications
Bayesia Inference Engine 1.1
A new release of Bayesia Inference Engine is now available. BIE 1.1 allows using approximate inference, constraint nodes, utility nodes and decision nodes. It is also possible to use the networks designed by Bayesia Modeling Engine
Bayesia Modeling Engine 1.0
A new Application Programing Interface (API) for Bayesian networks modeling is now available. Bayesia Modeling Engine allows constructing your Bayesian networks directly in your applications
Bayesia Graph Layouting Engine:
Powerful algorithms for the automatic layouting of your graphs
Bayesia Graph Layouting Engine 2.0
BGLE comes now with a new layouting algorithm. Besides the genetic algorithm necessary for the layouting of complex graphs, this new symmetric algorithm uses repulsion and attraction forces to efficiently compute the layouting of your graphs that are moderately connected