This paper resulted from postgraduate study and collaborative research. The paper describes a new approach to clustering problem based on geometric modeling of clusters and visualization of the clustering process. Clusters are described as geometric solid objects with function representations. Our approach applies the concepts of geometric modelling to clustering and uses data density as clustering criteria. An implementation of the proposed geometric clustering model is discussed. Visual clustering allows the user to see the result of clustering on the screen and set the appropriate parameters. The user visually clusters the data through graphics interface accessing dynamically 3-dimensional projections of multidimensional points from database or files.