- The Cumulative Sum (CUSUM) can be applied to detect some potential signal shifts (Carslaw et al., 2006; Andersen et al., 2009).
- A CUSUM procedure is developed by plotting the accumulated residuals between a measured variable and is used to test whether there is evidence that the residual time series is nonstationary or inhomogeneous.
- The change point can be calculated by interpretation of CUSUM chart.
- However, bootstrap analysis is done to ascertain the confidence level to quantify the mostly likely timing of change and to estimate the uncertainty associated with a change point.
- The following link gives information about CUSUM and bootstrap process: http://www.variation.com/cpa/tech/changepoint.html
- I performed bootstrap in MATLAB using the command: [bootstat,bootsam] = bootstrp(...) which was used as [bootstat,bootsam] = bootstrp(1000,@mean,A) to generate 1000 resamples with mean function for the data vector [A].
- But the confidence level estimated from these 1000 samples (using the example data set in above link) was found to be very low! That's what is making me confused regarding application of CUSUM method. Please if anyone can help me out!
Cyril Afuwai Gwazah
check
https://www.google.com.ng/url?q=https://arxiv.org/pdf/0906.1421&sa=U&ved=2ahUKEwjyqeOEtpvZAhWD_aQKHSqoBoIQFjAAegQIExAB&usg=AOvVaw2hqj6-9xSupNHFrvpkciyt
0 votes
0 thanks
Mohammad Zakwan
you may look read the paper on following link for easy bootstrapping process
//link.springer.com/article/10.1186/s40068-015-0031-4
1 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Geoscience
- Hydrology
More Priyank J. Sharma's questions See All
Similar questions and discussions