Time series are sequences of data collected over time. They are used in a wide range of applications, from weather forecasting to market analysis. However, time series can undergo changes in their stationarity, which means that their statistical properties cease to be constant over time.
The invention is a non-parametric method for detecting changes of stationarity in time series. The method is based on comparing the cumulative distributions of two segments of the series, using the Kolmogorov-Smirnov statistic.
A pointer is positioned at the start of the series. The Kolmogorov-Smirnov statistic is calculated to compare the data segments before and after the pointer. If the Kolmogorov-Smirnov statistic is greater than a critical value, the pointer is moved to the next point in the series. The process is repeated until the pointer reaches the end of the series.
Patent title:
System, apparatus and method for detecting stationarity changes in time series
Deposit Number:
BR 10 2012 009990 0
Pontifical Catholic University of Rio de Janeiro – PUC-Rio
Rua Marquês de São Vicente, 225, Gávea
Rio de Janeiro, RJ – Brasil
Zip code: 22451-900
Postal box: 38097
Phone:
+55 (21) 3527-2155