Data mining is a process of extracting data to get knowledge from large data repositories. One of data mining methods called classificatin, seeks classification model that is able to differentiate classes labels. Bayesian Network method is one of them. Bayesian Network consist of two components: the DAG structure depiting casuality relationship among data attributes and a containing tables of conditional probability based on previos attributes which is called CPT. CB* algorithm, which combines the two approaches, dependency analysis and search scoring, is one algorithm for constructing bayesian network structure from incomplete databases with missing values. This algorithm consist of two phases with phase one is to produce node ordering and phase two is construct DAG structure of bayesian network. The objective of this research is to analyse CB* algorithm from its function point of view which is reported to able to generate node ordering for producing structure which is markov equivalent to the original structure, and able to construct bayesian network from incomplete databases. The amount of missing values has no influence to the bayesian network structure. The research also includes the analysis of the capability of the algorithm to construct bayesian network structure whithout prior information. Based on the result, it is proven that the algorithm is capable to complete the trere tasks mentioned above. Keywords: Data Mining, Classification, Bayesian Network, Missing value, Node ordering