Enhancing Parallel Coordinates Visualization Using Genetic Algorithm with Smart Mutation

Visualization techniques have received a lot of attention regarding their potential to interpret and analyze the data.One of the marked visualization methods is the Parallel Coordinates Plot (PCP) utilized to high-dimensional datasets (more than three dimensions). Due to that, in visualizing large-scale datasets, the method suffers from high clutters produced from numerous intersection lines between neigh-boring axes, numbers of researchers have conducted techniques to boost PCPs. For instance, reducing the number of crossing lines by utilizing the re-ordering the neighboring axes in the PCP technique is a useful procedure to reduce the clutter. Motivated by this goal, the acquisition of the optimal coordinate’s order can be classified as a combinatorial optimization problem. However, in high-dimensional datasets, the optimization algorithm may face difficulty to deal with this issue. In this paper, we propose a smart mutation operator to enhance the performance of Genetic Algorithm (GA) in finding the optimal order of PCP based on diminishing the numerous intersection lines. However, any other user-desired metric can be utilized as an objective function. To assess the introduced method, we conducted a Monte Carlo simulation and several experiments to find an optimal coordinates’ order in PCP to visualize the datasets with various numbers of samples and dimensions. In the experimental results, utilizing the smart mutation represents an improvement in PCP visualization in terms of reducing the intersection lines between the neighboring coordinates compared to the original GA.