BIG DATA VISUALIZATION TOOLKIT
BiDViT is a useful tool for data visualization in many industries, including banking, manufacturing, life sciences, insurance, and advertising.
The human eye excels at perceiving patterns, outliers, and meaning in data based on its visual appearance but it becomes difficult to perceive such patterns when the data is presented in a dense and unorganized fashion, which is the case in most practical datasets that can have upwards of tens of thousands of data points.There has been an increasing need for algorithms and techniques for effectively visualizing/analyzing the large datasets of “big data”. BiDViT addresses the problem of clustering and visualizing such large datasets by effectively reducing the number of data points (i.e., the data size) such that the resulting clustered dataset can be clearly viewed and analyzed.
As a result of data clustering, similar data points can be grouped together and depicted using a single representative point. An analyst can then select representative points of interest and drill down to examine the constituent data points to determine patterns. BiDViT also lends its user an opportunity to compare the selected points against the entire dataset in general, on various attributes of the multidimensional datasets.
BiDViT is the only method that is designed to give a variety of clustering levels in a single run, thereby enabling an analyst to work with different levels of abstraction with a much finer level of control.
Performing data clustering on a large dataset (one million down to a few hundred thousand points) can take hours or more using traditional methods like k-means or spectral clustering, whereas BiDViT involves lower complexity and can be completed in only a few minutes. It is also massively parallelizable, meaning that this method can become even faster through parallel processing.
See BiDViT in action with these industry-applicable datasets. While BiDViT is able to visualize data for any industry, this demonstration includes examples from banking, automotive, and steel manufacturing. Each example involves various numbers of points and a variety of features related to each industry.
Which factors predict certain types of faults in steel manufacturing?