Meaning
Degrees of freedom provide the limit against which the complexity of a mathematical space is measured when mapping a dynamical system. Calculating the embedding dimension allows a founder to understand the complexity of the market feedback loops they are observing. The value is a cardinal integer that represents the minimum coordinates needed to describe the behavior of the system.
It stops applying when the system transitions from a deterministic state to one of pure random noise.
Manifold Complexity
Dimensionality reduction techniques seek to find the simplest representation of a complex data set without losing critical information. When the embedding dimension is high, many independent factors influence the outcome of a business decision. A low number indicates a more predictable environment where a few key variables dominate the landscape.
The measurement guides the selection of the appropriate analytical tools for the job.
Projection Depth
Mapping a time series into a higher dimensional space reveals structures that are invisible in the original one dimensional data. If the embedding dimension is incorrectly estimated, the resulting model may either oversimplify the reality or capture irrelevant noise. Proper estimation requires looking at how the density of points in the phase space changes as more dimensions are added.
This process continues until the trajectories of the system no longer intersect.
Variable Requirement
Resource allocation for data collection depends on knowing how many different types of measurements are necessary to monitor a system. Identifying the embedding dimension prevents the waste of energy on tracking variables that do not contribute to the understanding of the state. It ensures that the model is both efficient and robust against small fluctuations in individual inputs.
Complexity in the mapping determines the computational cost of the analysis.
