Geometric Data Analytics, Inc.

Expertise: Pattern & Anomaly Discovery

Detected patterns in QRS data.

Finding patterns in your data can be one of the most important approaches to analysis. Just as for modeling, pattern discovery can be approached using statistical methods or geometric ones. We specialize in combining the two methods to find the best solution.

Time Series occur all around us. Physicians measure pulse, respiration, blood pressure. Physiologists collect EKG, EEG, skin conductance. Both then look for patterns that predict vitality and health. Music contains many complex patterns that music analysts seek to extract. Video analysis is even more intricate and again rests on identifying patterns and characterizing how they change with time. Financial analysts confront some of the most complex time series data and much make predictions as fast as possible.

We use new methods to analyze time series data, based on Geometric and Topological Data Analysis to find patterns in time series. Complex rhythms can be detected automatically in music, regular movements can found in video. Patterns in EKG data are detected more naturally with GDA methods than using Fourier or Wavelet based methods because the only assumption we make is that a pattern is something that repeats (and not necessarily at regular intervals). At no time do we need to assume that the pattern has any particular form.

Higher Dimensional Data is of course more difficult to understand than data that we can visualize.  We use summary statistics based on shape to capture patterns in high dimensional data.