SûretéGlobale.Org and BAYESIA present a joint offer aver the crime statistics operating, BAYESIALAB CRIME ANALYST. Thanks to its mastery of advanced technology, Baysian networks, BayesiaLab helps you make better decisions. BayesiaLab models your expertise and turns your data into knowledge.

A bayesian network is a graphic probabilistic model through which one can acquire, capitalize on and exploit knowledge. Bayesian networks are the natural successors and heirs to symbolic, connectionist and statistical approaches to Artificial Intelligence and Data Mining.

They combine the rigour of powerful and stable mathematic formalism and the effectiveness of a «distributed» representation of knowledge and the readability of rule-based models. Particularly suited to taking uncertainty into consideration, they can as easily be described manually by experts in the field as they can be generated automatically through learning.

Until now Baysian networks have been exclusively used by high-level marketing analysis or diagnosis and security systems (EDF, PSA, etc. ..) or in spam filters or applications to profiling. BayesiaLab Crime Analyst is a dedicated version to the crime data analysis, achieved through the partnership between Baysia and SûretéGlobale.Org.

See the standard presentation of the software: here

Analyse graphique du noeud cibleNetwork modelling: capitalize on your expertise and quantify your uncertainties

A bayesian network is used to represent knowledge from a system (technical, computer, economic, biological, sociological, etc.) or to find out this knowledge by analysing data (learning).

Through the network one can then:

Thanks to its ergonomic interface, BayesiaLab enables you to easily formalize your knowledge in the form of bayesian networks.

Learning/data mining: discover the knowledge buried in your data Analyse des modalités du noeud cible



From your data on crime, BAYESIA CRIME ANALYST automatically creates the corresponding network and the probability tables of variables.

This is simply to answer the following question: "if variable A to the value VA, what is the probability that the variable B is another VB?"

Another singularity of Bayesian networks is being able to spread from one observation. Just force a variable to a value to see the probabilities in this context of all the other variables.

It is therefore easily noticeable causal relations, and we can take appropriate measures or test these measures on the system.

Questionnaire adaptatif : propositions en fonction de la pertinence (information/coût) de chaque variableThese properties have many predictive applications: “what is going to happen this morning?” or “Where is there the most pickpocketing? when? What's going to happen tomorrow in such or such street? etc. These functions are quasi-predictive, transitive and recursive, and thus allow to better focus the efforts of the Police forces, with an efficiency and a simplicity unprecedented.


Apprentissage de politique : où faire les sondages et où décider de forer ?Even better, Bayesian networks automatically learn from your data cause and effect relationships, and within the variables. It is therefore possible to answer questions like "If there were fewer robbery, what would happen? " or "what kind of crime should I concentrate to reduce overall crime in the territory?"


download case study


Analyse graphique du noeud cible