Data has shape, and that shape has meaning. Even though we typically do not think about data analysis in terms of shape, we do it complicity in our day-to-day practice. Even though being very typical in statistical analysis, linear shapes are not sufficient to describe more complicated and challenging data sets. There are various junctions in the trends, cyclic trends and many more reasons why a simple one to one linear relationship would fail. It is clear that by restricting ourselves to any predefined family of shapes, linear or non linear, we are searching for something in the data that is likely to lead to artificial results. This is implicit to data and does not learn anything from the purpose to model.
Topological Data Analysis (TDA) creates a descriptive model that adapts to the data while making minimal assumptions. In this workshops, by first quantifying and visualising the shape and global features of the underlying data, we aim to demonstrate the power of the TDA integrated into traditional data analysis and machine-learning techniques, achieving augmented superior results in various industrial problem-solving exercises, eg, risk management and banking..