Design space exploration, that is an embodiment of a paradigm Design by Shopping, refers to the systematic activity of discovery and evaluation of the elements in the design space in order to identify optimal solution by reducing the design space toward an area of performance. This activity is composed of three main phases: the discovery, the optimization and the selection. There are existing tools for the design space exploration with different graphs (ScatterPlot matrix, 2D and 3D scatter plot, parallel coordinates plot, etc.). These graphs are useful for the representation of multidimensional set of data with an unlimited number of alternatives (design points). Obviously, during the selection phase, designers face to a reduced design space with a limited number of design points (in a performance area). Thus, in our work, we try to identify which graph is the most adapted to the selection phase. It emerges, from literature, three graphs useful for the representation of multidimensional set of data (>3 variables) and with a limited number of alternatives (<50). Thus we have designed experimentation composed of 3 scenarios (with 13 design parameters and 5 variables of performance) performed by 30 participants. It results one graph more suited to the selection phase in the Design by Shopping: the Parallel Coordinates Plot.