Step 1: Choose your data (*.csv)

• The first column needs to contain ids and will not be visualized.
• All values in the data are considered to be on a ratio scale.
• Computing time increases with larger number of dimensions (depending on the ordering).

Download example datasets: Cars.csv , Synth. with 2 clusters , Synth. with 3 clusters , Synth. with 4 clusters

The values in all columns are in range [0,1] (no normalization is applied).

Step 2: Choose Ordering Methods

Default Ordering (same as input file) Clutter-based (Peng et al. 2004) Clutter-based Method for SG Contribution-based (Lu et al. 2015) Similarity-based (Artero et al. 2006) Disimilarity-based Method based on Artero et al. Similarity-based (Lu et al. 2015) Similarity-based; exhaustive search (Ankerst et al. 1998) Dissimilarity-based; exhaustive search (Blumenschein et al. 2020) Maximizing Variance of Neighboring Dimensions (Blumenschein et al. 2020) Similarity-based Method(Min Variance) Pattern Optimization Method (Makwana et al. 2012) Pargnostics (Dasgupta & Kosara 2010)

Euclidean Distance Pearson's Correlation Coefficient

Number of Line Crossings:


Angles of Crossing:


Parallelism:


Mutual Information:


Convergence, Divergence:


Over-plotting:

Monotonicity:


Symmetry:

Step 3: Run Visualization

Show data table with reordered dimensions

Download dataset with reordered dimensions as .csv file

Store similarity values of Max. Variance, Similarity, and Dissimilarity patterns

Visualize Parallel Coordinates