An Efficient Algorithm for Generating Interactive Visualization of Large Datasets

Abstract

Easy-to-use visualisation tools are essential for identifying patterns in unstructured data as it grows larger and more complicated. The majority of large data visualisation techniques either knowledge of a programming language or access to commercial tools. These days, data is growing in size, and statistical visualisation of the data needs numerous displays to represent all of the data's viewpoints. Model tuning, optimisation, and quick decision-making are greatly aided by the intriguing patterns produced by the models that are visualised. However, traditional techniques like histograms, pie charts, box plots, and bar graphs are typically insufficient to adequately communicate the intriguing pattern that needs to be extracted from massive datasets. As a result, this paper introduces a tree-plot method that uses the h2o package's in-memory node technique to store the enormous dataset in memory. Data. Tree's charting capabilities were employed to display the modelled trees, which were built using an Ada Boosted Model as the underlying learning methods. Visualising the modelled tree and evaluating the effectiveness of the data training process are based on the AUC, RMSE, execution procedure time, and MSE results.