From the daily values of stock indices to the minutely recorded number of your heart beats: time series, i.e. timely-ordered sequences of correlated observations, appear everywhere. Their analysis becomes increasingly important due to the massive production of data through, e.g. the internet of things or the digitalization of healthcare.
My research focuses on robust methods for time series analysis, i.e. on statistical techniques that are designed to handle data that contains outliers, structural changes, or other types of deviations from the expected patterns. Traditional time series analysis methods can be sensitive to such deviations and may produce inaccurate results. Robust methods aim to overcome this limitation by identifying and downweighting the influence of anomalies, while still preserving the overall structure of the time series. Ordinal pattern, which represent the spatial ordering of consecutive values in a time series, constitute one example of such methods.
People
