The visualisation, filter and reasoning of uncertain data has been a common concern for different areas such as industrial, military, commercial and medical diagnosis. For this reason, several researches have been focused on studying the called Bayesian Networks (BN) through the analysis of ideas about artificial intelligent, decision analysis and statistic communities. BN is a statistical modelling method used to represent uncertain causal relations between different statistical variables (a variable is a collectively exhaustive and mutually exclusive values) or propositions (Stutz, J. & P.Cheeseman, 1994).  In simpler terms a BN can be defined as a model or tool that can be used to model different things or scenarios e.g. the weather, a military battalion, a garbage disposal, either a disease or its symptoms (or both) for different reasons like encoding, learning and reasoning about probabilistic relationships. BN are very useful when the information about the past or the current circumstances are unclear, unavailable, uncertain, conflicting, indefinable incomplete and unpredictability or imprecise.

BN are very efficient in filtering, purifying and visualising the uncertain data with the help of various basic axioms, probabilistic calculus and, very efficient and effective algorithms that are used on different stages of this process for different purposes. This probabilistic calculus permits to capture the probabilistic relationship among the propositions or variables and also historical information about their relationships. The effects of BN tool vary with the situations; in some situations they may be more effective than the other. They are very effective when there is already some information and incoming data is uncertain or unavailable. Additionally, they are consistent for representing reasoning and effects through an intuitive graph. In the last year, the usage of BN has increased in the domain of automated reasoning such as data fusion and intelligent decision aids. The graphical framework of problem is also very useful feature of BN as it is very useful for computer and people (Finn V. Jensen, 1996).

If we compare BN with other related techniques we will find it is more useful from other techniques in different ways. If we compared them with knowledge based system, we will find the following advantages of BN over knowledge based system.First advantage of BN over knowledge based system its modular representation of uncertain knowledge, which makes them easier to maintain and to adapt to different contexts and second one is more intuitive knowledge representation for domain experts, making it easier for them to be involved in maintaining a system.

If we compared BN with neural network (i.e. is another technique for information processing), we will examine the following advantages of BN over neural network: Experts have the facility to provide knowledge in the form of causal structures, BN is more understandable and extensible as compare to neural network and BN can be used easily with missing data while it is not the case in neural networks.

Uncertainty can arise in different areas and in different situations. For example: uncertain to an expert about his/her knowledge, for inheritance in the situation being model, about accuracy or availability of information about a particular domain. As BN is very consist in representing uncertainty. Purpose of BN is to give estimates of certainties for events that are also not observable at an unacceptable cost.  The interaction among various reasoning and effects can be represented by using intuitive graph of BN. Actually they are very effective to model uncertain situation that depends upon reasoning and effects. BN can be applied in different fields of life e.g. medical diagnostic systems, weapons scheduling, processor fault diagnosis, generator monitoring expert system and troubleshooting.

Figure 2 shows a BN diagram for cancer disease. It can be interpreted as the conditional probabilities of getting cancer given the independent events of age and gender.