back to list

Project: Detection of similarities and correlations in multidimensional time series


Correlations are extensively used in all data-intensive disciplines, to identify relations between the data (e.g., relations between stocks, or between medical conditions and genetic factors). Most algorithms consider one-dimensional time series. For example, in the context of finance, the time series might represent the closing price of each stock, or the number of daily transactions of each stock, but not both. 

In many domains, it is critical to find correlations in ‘groups of time series’ together. Consider for example the case of meteorology sensors. Each sensor in this case may measure tens of different time series for the location (temperature, sea level pressure, wind speed, rainfall,and others). In other words, for each location we have a multi-dimensional time series, where the dimensions correspond to temperature, sea level pressure, etc.). To be able to say that the weather situation in two different locations  (say, location A and location B) exhibits some correlation, we need to check and compare these multi-dimensional time series as a whole, and not individual time series for the two locations.

In this thesis you may examine one of the following problems

  • Algorithms for efficiently detecting correlations in multidimensional time series

  • Algorithms for efficiently finding two or more multidimensional time series that are nearly identical (called near-duplicate detection)

  • Methods for summarizing multi-dimensional time series (e.g., with sketches or samples)

Prerequisites: ability to write efficient code in *Java or Scala*, comfortable with mathematical proofs, ability to read and understand scientific literature (conference papers and journal articles), successful completion of 2AMD15 with a high grade.

Odysseas Papapetrou
Get in contact