| Title: | Identify Breakpoints in Series of Data |
|---|---|
| Description: | Compute Buishand Range Test, Pettit Test, SNHT, Student t-test, and Mann-Whitney Rank Test, to identify breakpoints in series. For all functions NA is allowed. Since all of the mention methods identify only one breakpoint in a series, a general function to look for N breakpoint is given. Also, the Yamamoto test for climate jump is available. Alexandersson, H. (1986) <doi:10.1002/joc.3370060607>, Buishand, T. (1982) <doi:10.1016/0022-1694(82)90066-X>, Hurtado, S. I., Zaninelli, P. G., & Agosta, E. A. (2020) <doi:10.1016/j.atmosres.2020.104955>, Mann, H. B., Whitney, D. R. (1947) <doi:10.1214/aoms/1177730491>, Pettitt, A. N. (1979) <doi:10.2307/2346729>, Ruxton, G. D., jul (2006) <doi:10.1093/beheco/ark016>, Yamamoto, R., Iwashima, T., Kazadi, S. N., & Hoshiai, M. (1985) <doi:10.2151/jmsj1965.63.6_1157>. |
| Authors: | Santiago I. Hurtado [cre, aut] <[email protected]> |
| Maintainer: | Santiago I. Hurtado <[email protected]> |
| License: | GPL-3 |
| Version: | 1.2 |
| Built: | 2024-10-31 06:23:37 UTC |
| Source: | CRAN |
Look for several breakpoints
N_break_point(serie, n_max = 1, n_period=10, seed=FALSE, auto_select = FALSE, alpha = NULL,method='SNHT',dstr='norm', seed_set = 9658, change_random_seed = TRUE, seed_method = 6842)N_break_point(serie, n_max = 1, n_period=10, seed=FALSE, auto_select = FALSE, alpha = NULL,method='SNHT',dstr='norm', seed_set = 9658, change_random_seed = TRUE, seed_method = 6842)
serie |
numeric vector where the breakpoint is looked for |
n_max |
integer up to how many breakpoint should be test |
n_period |
an integer specifying the minimal length of a complete period to consider |
seed |
if FALSE (default) the first breakpoints to inicialize the iteration are calculated by splitting the serie in equal separated parts. If seed is given must be a list() of length n_max where it is specified the first breakpoints to take for each iteration. |
method |
which method should be used for breakpoint detection, supported: 'student', 'mann-whitney','SNHT' (default),'buishand' and 'pettit' |
dstr |
character specifying which distribution should be used for test simulations, only used if method is SNHT or buishand. Posible distributions 'norm' (default, normal dist), 'gamma', and 'self' (bootstrap) |
auto_select |
logical, should an automatic selection of how many breakpoints are be made, default FALSE |
alpha |
numeric, critical p value to use for auto_select |
seed_set |
Either a number to used to set a seed or NULL to set no seed inside the function |
change_random_seed |
Logical, can the .Random.seed change inside the function, or must remain the same after applying the function |
seed_method |
Either a number to used to set seed inside SNHT or buishand methods or NULL to set no seed |
Compute homogeneity test for all possible breaks in the serie considering several breakpoints. NA values are allow. In order to guarantee same result for the same input seed_set and seed_method (if method in SNHT or buishand) must be given.
N_break_point returns a list with the breakpoints index, it's p value and how many breakpoints are. If auto_select = F, a list with one list as specify for each n breakpoint tried
index where the breakpoints are found
p value of the test
how many breakpoints are looked for
Hurtado, S. I., Zaninelli, P. G., & Agosta, E. A. (2020). A multi-breakpoint methodology to detect changes in climatic time series. An application to wet season precipitation in subtropical Argentina. Atmospheric Research, 104955.
# Make a serie with three jumps, same as yamamoto example set.seed(524) x <- c(rnorm(30,1,1),rnorm(30,2,1),rnorm(30,1,1),rnorm(20,2,1)) # Look up to 5 breaks using pettit break_prosition <- N_break_point(serie=x, n_max = 5, method='pettit', auto_select=TRUE,alpha=0.1) plot(x) abline(v = break_prosition$breaks, col='red')# Make a serie with three jumps, same as yamamoto example set.seed(524) x <- c(rnorm(30,1,1),rnorm(30,2,1),rnorm(30,1,1),rnorm(20,2,1)) # Look up to 5 breaks using pettit break_prosition <- N_break_point(serie=x, n_max = 5, method='pettit', auto_select=TRUE,alpha=0.1) plot(x) abline(v = break_prosition$breaks, col='red')
Compute Buishand Range Test or Standard Normal Homogeneity Test for a serie, NAs allow in both Test
SNHT(serie,n_period=10,dstr='norm',simulations = 1000, seed_set = 9658, change_random_seed = TRUE) Buishand_R(serie,n_period=10,dstr='norm',simulations = 1000, seed_set = 9658, change_random_seed = TRUE)SNHT(serie,n_period=10,dstr='norm',simulations = 1000, seed_set = 9658, change_random_seed = TRUE) Buishand_R(serie,n_period=10,dstr='norm',simulations = 1000, seed_set = 9658, change_random_seed = TRUE)
serie |
numeric vector where the breakpoint is looked for |
n_period |
an integer specifying the minimal length of a complete period to consider |
dstr |
character specifying which distribution should be used for test simulations, 'norm' (default; normal distribution), 'gamma', and 'self' (will compute bootstrap) |
simulations |
an integer specifying how many Monte Carlo simulations to perform, default is 1000. |
seed_set |
Either a number to used to set a seed or NULL to set no seed inside the function |
change_random_seed |
Logical, can the .Random.seed change inside the function, or must remain the same after applying the function |
SNHT compute Standard Normal Homogeneity Test where NA values are allow. In order to guarantee same result for the same input seed_set must be given.
Buishand_R Compute Buishand Range Test for Homogeneity where NA values are allow. In order to guarantee same result for the same input seed_set must be given.
SNHT and Buishand_R returns a list with the breakpoint index and it's p value
index where the breakpoint is found
p value of the test
- Alexandersson, H., jan 1986. A homogeneity test applied to precipitation data. Journal of Climatology 6 (6), 661–675. URL http://doi.wiley.com/10.1002/joc.3370060607
- Buishand, T., aug 1982. Some methods for testing the homogeneity of rainfall records. Journal of Hydrology 58 (1-2), 11–27. URL https://doi.org/10.1016/0022-1694(82)90066-X
# Make a serie with one breakpoint x <- c(rnorm(60,1,1),rnorm(40,2,1)) # Look for break using SNHT, Buishand_R can be used in exactly the same way break_prosition <- SNHT(serie = x) plot(x) abline(v = break_prosition$breaks)# Make a serie with one breakpoint x <- c(rnorm(60,1,1),rnorm(40,2,1)) # Look for break using SNHT, Buishand_R can be used in exactly the same way break_prosition <- SNHT(serie = x) plot(x) abline(v = break_prosition$breaks)
Compute Rolling Mann-Whitney-Wilcoxon Test, Rolling Student t-test and Pettit test for homogeneity, NAs allow.
stu(serie,n_period=10) man.whi(serie,n_period=10) pettit(serie,n_period=10)stu(serie,n_period=10) man.whi(serie,n_period=10) pettit(serie,n_period=10)
serie |
numeric vector where the breakpoint is looked for |
n_period |
an integer specifying the minimal length of a complete period to consider |
man.whi compute Mann-Whitney-Wilcoxon Test and stu the Student t-test for all possible breaks in the serie and get the most significant break.In both test NA values are allow.
pettit Compute the Pettit Test for Homogeneity. NA values are allow.
pettit, man.whi and stu returns a list with the breakpoint index and its p value
index where the breakpoint is found
p value of the test
- Ruxton, G. D., jul 2006. The unequal variance t-test is an underused alternative to Student’s t-test and the Mann–Whitney U test. Behavioral Ecology 17 (4), 688–690. URL: http://academic.oup.com/beheco/article/17/4/688/215960/The-unequal-variance-ttest-is-an-underused
- Mann, H. B., Whitney, D. R., mar 1947. On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other. The Annals of Mathematical Statistics 18 (1), 50–60. URL http://projecteuclid.org/euclid.aoms/1177730491
- Pettitt, A. N., 1979. A Non-Parametric Approach to the Change-Point Problem. Applied Statistics 28 (2), 126. URL https://www.jstor.org/stable/10.2307/2346729?origin=crossref
# Make a serie with one breakpoint x <- c(rnorm(60,1,1),rnorm(40,2,1)) # Look for break using pettit(), man.whi() and stu() break_prosition_pettit <- pettit(serie = x) break_prosition_man.whi <- man.whi(serie = x) break_prosition_stu <- stu(serie = x) plot(x) abline(v = break_prosition_pettit$breaks,col='red') abline(v = break_prosition_man.whi$breaks,col='blue') abline(v = break_prosition_stu$breaks,col= 'green')# Make a serie with one breakpoint x <- c(rnorm(60,1,1),rnorm(40,2,1)) # Look for break using pettit(), man.whi() and stu() break_prosition_pettit <- pettit(serie = x) break_prosition_man.whi <- man.whi(serie = x) break_prosition_stu <- stu(serie = x) plot(x) abline(v = break_prosition_pettit$breaks,col='red') abline(v = break_prosition_man.whi$breaks,col='blue') abline(v = break_prosition_stu$breaks,col= 'green')
Compute Yamamoto Test for climate jumps in a serie.
yamamoto(serie, alpha = 0.1, n_period = 10)yamamoto(serie, alpha = 0.1, n_period = 10)
serie |
numeric vector where the breakpoint is looked for |
n_period |
an integer specifying the length of the window to use, can not bet odd |
alpha |
numeric, p value to use |
yamamoto compute the Yamamoto Test.
yamamoto returns a list with the breakpoints indexes and the amount
vector of indexes where the breakpoint is found
Amount of breakpoints
Yamamoto, R., Iwashima, T., Kazadi, S. N., & Hoshiai, M. (1985). Climatic jump: a hypothesis in climate diagnosis. Journal of the Meteorological Society of Japan. Ser. II, 63(6), 1157-1160.
# Make a serie with three jumps, same as N_break_point example set.seed(524) x <- c(rnorm(30,1,1),rnorm(30,2,1),rnorm(30,1,1),rnorm(20,2,1)) # Look for break using yamamoto() break_prosition <- yamamoto(serie = x) plot(x) abline(v = break_prosition$breaks, col='red')# Make a serie with three jumps, same as N_break_point example set.seed(524) x <- c(rnorm(30,1,1),rnorm(30,2,1),rnorm(30,1,1),rnorm(20,2,1)) # Look for break using yamamoto() break_prosition <- yamamoto(serie = x) plot(x) abline(v = break_prosition$breaks, col='red')