PCDC - On the Highway to Data - A Tool for the Fast Generation of Large Synthetic Data Sets

Abstract

In this paper, we present Parallel Coordinates for Data Creation (PCDC), a new visual-interactive method for the fast generation of labeled multidimensional data sets. Multivariate data need to be analyzed in various domains such as finance, biology or medicine using complex data mining techniques. For the evaluation or presentation of the techniques, e.g., for assessing their sensitivity to specific data properties, test data need to be generated. PCDC allows for a fast and intuitive creation of multivariate data with several classes. It is based on interactive definition of data regions and data distributions in a parallel coordinates view. It offers a quick definition of data regions over several dimensions in one interface. Moreover, the users can directly see the outcome of their settings in the same view without the need for switching between data generation and output visualization. Our tool enables also an easy adjustment of the data generation parameters for creating additional similar datasets.