4.1.3. Projection of Vectors

On the basis of the composite distance matrix, we map samples and attributes into visual space. In contrast to the data sets in the data context map [25], air quality data are characterized by having a sample number that is much larger than the number of attributes, and *VV* stores less information than *DD* and *DV*. To ensure that the accuracy of *DD* and *DV* is maintained as much as possible in the projection layout, we adopt two mapping steps: we map the samples first and then embed the attributes into the samples layout. Figure 3 shows the pipeline of the whole projection process.
