New method PANIC detects pervasive nonstationarity in large datasets.
A new method called PANIC was developed to analyze data with nonstationarity. PANIC can tell if the nonstationarity in data is widespread or specific to certain parts. It can also find out how many different trends are affecting the data. PANIC tests the hidden parts of the data, not just what we can see. It can estimate these hidden factors accurately, even if we don't know if they are stable or changing. The tests done with PANIC have been shown to work well with real data.