perSimplex is an intelligent software tool designed for numeric data processing aimed to obtain valuable knowledge and information. perSimplex is designed for everybody who is intending to evaluate big amount of data stored in computer databases in order to support the decision making and management tasks. The standard need of the users, who represent the numeric data and their changes in the traditional curve diagram, gave us the basic idea. The graphic representation is advantageous due to several reasons.
The main one is better perception and understandability of the graphic representation than that of tables full of numbers and dates. Simultaneously, the shape of the diagram curves is important since it is just this shape, which informs the user on the movement and changes of the numeric data.
For the user, it is crucial how deep is the curve slope while decreasing or increasing as well as it's any specific shape. The problem arises as soon as the user has placed too much curves in the diagram. As shown in the figure, the diagram becomes less apprehensible even for as few as 100 different curves. Therefore, if you want represent graphically thousands or millions of curves, such diagram becomes useless for practical purposes.
The solution of this problem is simple: The diagram curves should be divided in groups – clusters of curves with similar shape. And this is the basis of perSimplex. We just would like to remind you, that it is just the curve shape, which motivates the users to use the curve diagram for the data representation. Therefore, the shape similarity for the curves is a natural criterion for creation of clusters.
The input data shall not necessarily express only the dynamics change in time, i.e. the situation when the individual columns identify the daily hours, month days or year weeks. The data may include data which is not homogeneous, e.g. sex, age, qualification, income or number of children. The only thing needed is the sufficient numerical representation of the data and then the use of perSimplex. Then, the cluster generation criterion becomes multidimensional. This allows the dealing with tasks related to the factor analysis.
Please note that for real data processing, the curves in the identified clusters have only approximately identical shape instead of the absolutely identical one.
dynamics change | natural structure | graph | imprecision | human thinking | technology | fuzzy sets | Max Black | valuated output | mathematical method | computer databases | crisp logic | artificial | exact result | nonlinear data intelligence | ordered output | Lotfi A. Zadeh | expert systems | thinking process | human tolerance | decision support systems | uncertainty | ranking | warehouse | Bart Kosko | vagueness | approximating | Earl Cox | query | ambiguous data | imprecise data | degree of membership | data identifying | information overload | intelligent information retrieval | factor analysis | large databases | multivalued logic | crisp set | numerical representation | data representation | multidimensional | software tool | knowledge | data warehousing | fuzzy evaluation | approximate data | information processing | information | graphic representation | data mining | fuzzy logic | relevancy valuation | identification | relevant information | fuzzy databases |
