Discover hidden structures in data

Empower data exploration and improve model performance for complex data with Topological Data Analysis (TDA)

Goal

Analyzing rich and big financial data with the aim of discovering “hidden structures” is a challenging task. When only two or three aspects are of interest, that information can be easily placed on a coordinate system for gaining simple insights, but with potentially dozens of data dimensions a new approach is necessary. Smarter, better tools are required in order to further learn about the data, to visualize, to group and to create meaningful features in order to generate reliable predictions.

Insight & Action

Topological Data Analysis (TDA) is a yet little-known tool that allows to effectively work with high-dimensional data. In TDA, data is considered as a mathematical object and the aim is to characterize the “shape” of this object. TDA is able to render high dimensional data in a more comprehensible form.  One way to do this is by representing data as a “topological network”. This kind of network provides an insightful segmentation of the data that, in turn, offers the opportunity to identify relevant “topological features”.

Results

TDA is a powerful tool to obtain insights from complex techniques. By focusing on topological invariants, Knowledge Lab is able to discover subgroups in the data that with other methodologies stay undetected. Through TDA, we visualize complex data, gain deeper insights and find “hidden structures”. The result is a focus on the relevant aspects of data and therewith the creation of a higher business value.