diff --git a/R/poiss_boostrap_em.R b/R/poiss_boostrap_em.R
index ca20b322fcd2ca8e178d0f248447cf384698cc6e..82022c8471653ae81fc38548b5c6284c1dc8fff1 100644
--- a/R/poiss_boostrap_em.R
+++ b/R/poiss_boostrap_em.R
@@ -4,12 +4,14 @@
#'
#' @param x a matrix of male female kmers counts
#' @param params a list of init parameters
+#' @param gamma a vector of kappa prior
#' @param threshold theshold to stop the EM algorithm
+#' @param nbatch (default: 10) the number of batch
#' @param nboot number of boostrap sample
-#' @param frac (default: 1.) fraction of the data to use at each step
#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
#' @param core number of cpus to use for the computation
#' @param max_iter (default: 100) maximum number of iteration
+#' @param max_error (default: 5) maximum number of try
#' @return tibble of the statistics
#' @importFrom dplyr sample_n sample_frac
#' @importFrom parallel mclapply
@@ -20,73 +22,23 @@
#' extraDistr::rbvpois(100, 20, 0, 5) # Y
#' ) %>% pois_boostrap_EM(nboot = 10)
#' @noRd
-poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, frac = 1, bootsize = 1, core = 1, max_iter = 100) {
+poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, nbatch = nbatch, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
params = list(
A = list(kappa = 0.4, l1 = mean(x), l2 = mean(x), l3 = mean(x))
)
- if (sex %in% c("XO", "XY")) {
- params$X <- list(kappa = 0.3, l1 = 1, l2 = mean(x), l3 = mean(x))
- }
- if (sex == "XY") {
- params$Y <- list(kappa = 0.3, l1 = mean(x), l2 = 1, l3 = mean(x))
- }
+ gamma <- c(round(nrow(x) * bootsize))
+ if (sex %in% c("XO", "XY")) {
+ params$X <- list(kappa = 0.3, l1 = 1, l2 = mean(x), l3 = mean(x))
+ gamma <- c(round(nrow(x) * bootsize) * .8, round(nrow(x) * bootsize) * .2)
+ }
+ if (sex == "XY") {
+ params$Y <- list(kappa = 0.3, l1 = mean(x), l2 = 1, l3 = mean(x))
+ gamma <- c(round(nrow(x) * bootsize) * .8, round(nrow(x) * bootsize) * .15, round(nrow(x) * bootsize) * .05)
+ }
if (core == 1) {
- res <- lapply(as.list(1:nboot), function(iter, x, bootsize) {
- res <- x %>%
- as_tibble() %>%
- sample_frac(bootsize, replace = T) %>%
- as.matrix() %>%
- em_bipoiss_clust(params = params, frac = frac, max_iter = max_iter)
- print(tibble(loglik = res$loglik, BIC = res$BIC))
- return(tibble(loglik = res$loglik, BIC = res$BIC))
- }, x = x, bootsize = bootsize)
+ res <- lapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, nbatch = nbatch, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error)
} else {
- res <- parallel::mclapply(as.list(1:nboot), function(iter, x, bootsize) {
- res <- tryCatch({
- x %>%
- as_tibble() %>%
- sample_frac(bootsize, replace = T) %>%
- as.matrix() %>%
- em_bipoiss_clust(params, frac = frac, max_iter = max_iter)
- }, error = function(e) {
- x %>%
- as_tibble() %>%
- sample_frac(bootsize, replace = T) %>%
- as.matrix() %>%
- em_bipoiss_clust(params, frac = frac, max_iter = max_iter)
- })
- return(tibble(loglik = res$loglik, BIC = res$BIC))
- }, x = x, bootsize = bootsize, mc.cores = core)
+ res <- parallel::mclapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, nbatch = nbatch, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error, mc.cores = core)
}
return(res)
}
-
-#' Function to compile statistics differences between model XY and XO on boostrap
-#' samples of the data
-#'
-#' @param x a matrix of male female kmers counts
-#' @param threshold theshold to stop the EM algorithm
-#' @param nboot number of boostrap sample
-#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
-#' @param frac (default: 1.) fraction of the data to use at each step
-#' @param core number of cpus to use for the computation
-#' @param max_iter (default: 100) maximum number of iteration
-#' @return tibble with the statistics
-#' @importFrom tidyr pivot_longer
-#' @export
-#' @examples
-#' res <- rbind(
-#' extraDistr::rbvpois(1000, 10, 10, 20), # A
-#' extraDistr::rbvpois(1000, 0, 20, 20), # X
-#' extraDistr::rbvpois(1000, 20, 0, 5) # Y
-#' ) %>%
-#' poiss_compare_models(nboot = 3, core = 1)
-poiss_compare_models <- function(x, threshold = 1, nboot = 100, bootsize = 1, frac = 1, core = 1, max_iter = 100) {
- tibble(
- model_XY = poiss_boostrap_EM(x, sex = "XY", threshold = threshold, nboot = nboot, frac = 1, bootsize = bootsize, core = core, max_iter = max_iter),
- model_XO = poiss_boostrap_EM(x, sex = "XO", threshold = threshold, nboot = nboot, frac = 1, bootsize = bootsize, core = core, max_iter = max_iter),
- model_OO = poiss_boostrap_EM(x, sex = "OO", threshold = threshold, nboot = nboot, frac = 1, bootsize = bootsize, core = core, max_iter = max_iter)
- ) %>%
- pivot_longer(cols = c(model_XY, model_XO, model_OO)) %>%
- unnest(c(value))
-}
diff --git a/R/poiss_boostrap_em_sub.R b/R/poiss_boostrap_em_sub.R
index d59aed85fbd47de1dcb47ac6d89f42c423f50fb2..09f0ef91ba7f1299edc8558ab10831230c58a9cb 100644
--- a/R/poiss_boostrap_em_sub.R
+++ b/R/poiss_boostrap_em_sub.R
@@ -7,6 +7,7 @@
#' @param gamma a vector of kappa prior
#' @param threshold theshold to stop the EM algorithm
#' @param nboot number of boostrap sample
+#' @param nbatch (default: 10) the number of batch
#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
#' @param core number of cpus to use for the computation
#' @param max_iter (default: 100) maximum number of iteration
@@ -21,7 +22,7 @@
#' extraDistr::rbvpois(100, 20, 0, 5) # Y
#' ) %>% pois_boostrap_EM(nboot = 10)
#' @noRd
-poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, max_error) {
+poiss_boostrap_EM_sub <- function(iter, x, bootsize, nbatch = nbatch, params, gamma, max_iter, max_error) {
res <- list()
res$loglik <- c(NA)
iter <- 0
@@ -30,80 +31,8 @@ poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, ma
as_tibble() %>%
sample_frac(bootsize, replace = T) %>%
as.matrix() %>%
- em_bipoiss_clust(params = params, gamma = gamma, max_iter = max_iter)
+ em_bipoiss_clust(params = params, gamma = gamma, max_iter = max_iter, nbatch = nbatch)
iter <- iter + 1
}
return(tibble(loglik = res$loglik, BIC = res$BIC))
}
-
-#' Function to compile statistics bootstrap samples of the data
-#'
-#' @param x a matrix of male female kmers counts
-#' @param params a list of init parameters
-#' @param gamma a vector of kappa prior
-#' @param threshold theshold to stop the EM algorithm
-#' @param nboot number of boostrap sample
-#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
-#' @param core number of cpus to use for the computation
-#' @param max_iter (default: 100) maximum number of iteration
-#' @param max_error (default: 5) maximum number of try
-#' @return tibble of the statistics
-#' @importFrom dplyr sample_n sample_frac
-#' @importFrom parallel mclapply
-#' @examples
-#' rbind(
-#' extraDistr::rbvpois(1000, 10, 10, 20), # A
-#' extraDistr::rbvpois(500, 0, 20, 20), # X
-#' extraDistr::rbvpois(100, 20, 0, 5) # Y
-#' ) %>% pois_boostrap_EM(nboot = 10)
-#' @noRd
-poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
- params = list(
- A = list(kappa = 0.4, l1 = mean(x), l2 = mean(x), l3 = mean(x))
- )
- gamma <- c(round(nrow(x) * bootsize))
- if (sex %in% c("XO", "XY")) {
- params$X <- list(kappa = 0.3, l1 = 1, l2 = mean(x), l3 = mean(x))
- gamma <- c(round(nrow(x) * bootsize) * .8, round(nrow(x) * bootsize) * .2)
- }
- if (sex == "XY") {
- params$Y <- list(kappa = 0.3, l1 = mean(x), l2 = 1, l3 = mean(x))
- gamma <- c(round(nrow(x) * bootsize) * .8, round(nrow(x) * bootsize) * .15, round(nrow(x) * bootsize) * .05)
- }
- if (core == 1) {
- res <- lapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error)
- } else {
- res <- parallel::mclapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error, mc.cores = core)
- }
- return(res)
-}
-
-#' Function to compile statistics differences between model XY and XO on boostrap
-#' samples of the data
-#'
-#' @param x a matrix of male female kmers counts
-#' @param threshold theshold to stop the EM algorithm
-#' @param nboot number of boostrap sample
-#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
-#' @param core number of cpus to use for the computation
-#' @param max_iter (default: 100) maximum number of iteration
-#' @param max_error (default: 5) maximum number of try
-#' @return tibble with the statistics
-#' @importFrom tidyr pivot_longer
-#' @export
-#' @examples
-#' res <- rbind(
-#' extraDistr::rbvpois(1000, 10, 10, 20), # A
-#' extraDistr::rbvpois(1000, 0, 20, 20), # X
-#' extraDistr::rbvpois(1000, 20, 0, 5) # Y
-#' ) %>%
-#' poiss_compare_models(nboot = 3, core = 1)
-poiss_compare_models <- function(x, threshold = 1, nboot = 100, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
- tibble(
- model_XY = poiss_boostrap_EM(x, sex = "XY", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
- model_XO = poiss_boostrap_EM(x, sex = "XO", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
- model_OO = poiss_boostrap_EM(x, sex = "OO", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error)
- ) %>%
- pivot_longer(cols = c(model_XY, model_XO, model_OO)) %>%
- unnest(c(value))
-}
diff --git a/R/poiss_compare_models.R b/R/poiss_compare_models.R
new file mode 100644
index 0000000000000000000000000000000000000000..31b80336d139d405181a6ca3140bb4d81050466f
--- /dev/null
+++ b/R/poiss_compare_models.R
@@ -0,0 +1,32 @@
+# WARNING - Generated by {fusen} from dev/flat_full_poisson.Rmd: do not edit by hand
+
+#' Function to compile statistics differences between model XY and XO on boostrap
+#' samples of the data
+#'
+#' @param x a matrix of male female kmers counts
+#' @param threshold theshold to stop the EM algorithm
+#' @param nboot number of boostrap sample
+#' @param nbatch (default: 10) the number of batch
+#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
+#' @param core number of cpus to use for the computation
+#' @param max_iter (default: 100) maximum number of iteration
+#' @param max_error (default: 5) maximum number of try
+#' @return tibble with the statistics
+#' @importFrom tidyr pivot_longer
+#' @export
+#' @examples
+#' res <- rbind(
+#' extraDistr::rbvpois(1000, 10, 10, 20), # A
+#' extraDistr::rbvpois(1000, 0, 20, 20), # X
+#' extraDistr::rbvpois(1000, 20, 0, 5) # Y
+#' ) %>%
+#' poiss_compare_models(nboot = 3, core = 1)
+poiss_compare_models <- function(x, threshold = 1, nboot = 100, nbatch = 10, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
+ tibble(
+ model_XY = poiss_boostrap_EM(x, sex = "XY", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
+ model_XO = poiss_boostrap_EM(x, sex = "XO", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
+ model_OO = poiss_boostrap_EM(x, sex = "OO", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error)
+ ) %>%
+ pivot_longer(cols = c(model_XY, model_XO, model_OO)) %>%
+ unnest(c(value))
+}
diff --git a/dev/config_fusen.yaml b/dev/config_fusen.yaml
index f7a28e87a26900ce9de95f26f57ac10c3997d0b8..d84d1641adba95d2214e4302df0cff8806bd51b1 100644
--- a/dev/config_fusen.yaml
+++ b/dev/config_fusen.yaml
@@ -60,8 +60,12 @@ flat_full_poisson.Rmd:
- R/loglik_bipoiss_clust.R
- R/bic_bipoiss_clust.R
- R/param_diff_poiss.R
+ - R/get_cluster_poiss.R
+ - R/e_bipoiss_clust_batch.R
- R/em_bipoiss_clust.R
+ - R/poiss_boostrap_em_sub.R
- R/poiss_boostrap_em.R
+ - R/poiss_compare_models.R
tests: []
vignettes: vignettes/getting-started-poisson-version.Rmd
inflate:
diff --git a/dev/flat_full_poisson.Rmd b/dev/flat_full_poisson.Rmd
index fdfc0ba69e58f9ac5047c2fd8c597227876b2dac..858565cdf595252a01ffe736320384a762479659 100644
--- a/dev/flat_full_poisson.Rmd
+++ b/dev/flat_full_poisson.Rmd
@@ -679,6 +679,7 @@ extraDistr::rbvpois(1000*res[[1]]$kappa, res[[1]]$l1, res[[1]]$l2, res[[1]]$l3)
# Model comparaison
+## boostrap EM sub
```{r fun-poiss_boostrap_EM_sub}
#' Function to compile statistics bootstrap samples of the data
#'
@@ -687,6 +688,7 @@ extraDistr::rbvpois(1000*res[[1]]$kappa, res[[1]]$l1, res[[1]]$l2, res[[1]]$l3)
#' @param gamma a vector of kappa prior
#' @param threshold theshold to stop the EM algorithm
#' @param nboot number of boostrap sample
+#' @param nbatch (default: 10) the number of batch
#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
#' @param core number of cpus to use for the computation
#' @param max_iter (default: 100) maximum number of iteration
@@ -700,7 +702,7 @@ extraDistr::rbvpois(1000*res[[1]]$kappa, res[[1]]$l1, res[[1]]$l2, res[[1]]$l3)
#' extraDistr::rbvpois(500, 0, 20, 20), # X
#' extraDistr::rbvpois(100, 20, 0, 5) # Y
#' ) %>% pois_boostrap_EM(nboot = 10)
-poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, max_error) {
+poiss_boostrap_EM_sub <- function(iter, x, bootsize, nbatch = nbatch, params, gamma, max_iter, max_error) {
res <- list()
res$loglik <- c(NA)
iter <- 0
@@ -709,13 +711,14 @@ poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, ma
as_tibble() %>%
sample_frac(bootsize, replace = T) %>%
as.matrix() %>%
- em_bipoiss_clust(params = params, gamma = gamma, max_iter = max_iter)
+ em_bipoiss_clust(params = params, gamma = gamma, max_iter = max_iter, nbatch = nbatch)
iter <- iter + 1
}
return(tibble(loglik = res$loglik, BIC = res$BIC))
}
```
+## Bootstrap EM
```{r fun-poiss_boostrap_EM}
#' Function to compile statistics bootstrap samples of the data
#'
@@ -723,6 +726,7 @@ poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, ma
#' @param params a list of init parameters
#' @param gamma a vector of kappa prior
#' @param threshold theshold to stop the EM algorithm
+#' @param nbatch (default: 10) the number of batch
#' @param nboot number of boostrap sample
#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
#' @param core number of cpus to use for the computation
@@ -737,7 +741,7 @@ poiss_boostrap_EM_sub <- function(iter, x, bootsize, params, gamma, max_iter, ma
#' extraDistr::rbvpois(500, 0, 20, 20), # X
#' extraDistr::rbvpois(100, 20, 0, 5) # Y
#' ) %>% pois_boostrap_EM(nboot = 10)
-poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
+poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, nbatch = nbatch, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
params = list(
A = list(kappa = 0.4, l1 = mean(x), l2 = mean(x), l3 = mean(x))
)
@@ -751,14 +755,15 @@ poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, bootsiz
gamma <- c(round(nrow(x) * bootsize) * .8, round(nrow(x) * bootsize) * .15, round(nrow(x) * bootsize) * .05)
}
if (core == 1) {
- res <- lapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error)
+ res <- lapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, nbatch = nbatch, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error)
} else {
- res <- parallel::mclapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error, mc.cores = core)
+ res <- parallel::mclapply(as.list(1:nboot), FUN = poiss_boostrap_EM_sub, x = x, bootsize = bootsize, nbatch = nbatch, params = params, gamma = gamma, max_iter = max_iter, max_error = max_error, mc.cores = core)
}
return(res)
}
```
+# Model compare
```{r fun-poiss_compare_models}
#' Function to compile statistics differences between model XY and XO on boostrap
#' samples of the data
@@ -766,6 +771,7 @@ poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, bootsiz
#' @param x a matrix of male female kmers counts
#' @param threshold theshold to stop the EM algorithm
#' @param nboot number of boostrap sample
+#' @param nbatch (default: 10) the number of batch
#' @param bootsize size of the boostrap sample (if < 0 we take a percentage of x)
#' @param core number of cpus to use for the computation
#' @param max_iter (default: 100) maximum number of iteration
@@ -780,11 +786,11 @@ poiss_boostrap_EM <- function(x, sex = "XY", threshold = 1, nboot = 100, bootsiz
#' extraDistr::rbvpois(1000, 20, 0, 5) # Y
#' ) %>%
#' poiss_compare_models(nboot = 3, core = 1)
-poiss_compare_models <- function(x, threshold = 1, nboot = 100, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
+poiss_compare_models <- function(x, threshold = 1, nboot = 100, nbatch = 10, bootsize = 1, core = 1, max_iter = 100, max_error = 5) {
tibble(
- model_XY = poiss_boostrap_EM(x, sex = "XY", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
- model_XO = poiss_boostrap_EM(x, sex = "XO", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
- model_OO = poiss_boostrap_EM(x, sex = "OO", threshold = threshold, nboot = nboot, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error)
+ model_XY = poiss_boostrap_EM(x, sex = "XY", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
+ model_XO = poiss_boostrap_EM(x, sex = "XO", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error),
+ model_OO = poiss_boostrap_EM(x, sex = "OO", threshold = threshold, nboot = nboot, nbatch = nbatch, bootsize = bootsize, core = core, max_iter = max_iter, max_error = max_error)
) %>%
pivot_longer(cols = c(model_XY, model_XO, model_OO)) %>%
unnest(c(value))
@@ -801,7 +807,7 @@ data <- rbind(
extraDistr::rbvpois(n, 0, 200, 200), # X
extraDistr::rbvpois(n, 200, 0, 5) # Y
)
-res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +
@@ -819,7 +825,7 @@ data <- rbind(
extraDistr::rbvpois(n, 100, 100, 200), # A
extraDistr::rbvpois(n, 0, 200, 200) # X
)
-res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +
@@ -836,7 +842,7 @@ We simulate 1 clusters from bi-Poisson distribution
data <- rbind(
extraDistr::rbvpois(10000, 100, 100, 200) # A
)
-res <- data %>% poiss_compare_models(nboot = 10, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +
diff --git a/man/poiss_compare_models.Rd b/man/poiss_compare_models.Rd
index 852b3fd9242c6ae00b6af80543f3105e448706c8..9bd09992b06e4f0fc3daefbf96fd3aa7024e6b59 100644
--- a/man/poiss_compare_models.Rd
+++ b/man/poiss_compare_models.Rd
@@ -1,5 +1,5 @@
% Generated by roxygen2: do not edit by hand
-% Please edit documentation in R/poiss_boostrap_em.R, R/poiss_boostrap_em_sub.R
+% Please edit documentation in R/poiss_compare_models.R
\name{poiss_compare_models}
\alias{poiss_compare_models}
\title{Function to compile statistics differences between model XY and XO on boostrap
@@ -9,16 +9,7 @@ poiss_compare_models(
x,
threshold = 1,
nboot = 100,
- bootsize = 1,
- core = 1,
- max_iter = 100,
- max_error = 5
-)
-
-poiss_compare_models(
- x,
- threshold = 1,
- nboot = 100,
+ nbatch = 10,
bootsize = 1,
core = 1,
max_iter = 100,
@@ -32,6 +23,8 @@ poiss_compare_models(
\item{nboot}{number of boostrap sample}
+\item{nbatch}{(default: 10) the number of batch}
+
\item{bootsize}{size of the boostrap sample (if < 0 we take a percentage of x)}
\item{core}{number of cpus to use for the computation}
@@ -39,28 +32,15 @@ poiss_compare_models(
\item{max_iter}{(default: 100) maximum number of iteration}
\item{max_error}{(default: 5) maximum number of try}
-
-\item{frac}{(default: 1.) fraction of the data to use at each step}
}
\value{
-tibble with the statistics
-
tibble with the statistics
}
\description{
-Function to compile statistics differences between model XY and XO on boostrap
-samples of the data
-
Function to compile statistics differences between model XY and XO on boostrap
samples of the data
}
\examples{
-res <- rbind(
- extraDistr::rbvpois(1000, 10, 10, 20), # A
- extraDistr::rbvpois(1000, 0, 20, 20), # X
- extraDistr::rbvpois(1000, 20, 0, 5) # Y
- ) \%>\%
- poiss_compare_models(nboot = 3, core = 1)
res <- rbind(
extraDistr::rbvpois(1000, 10, 10, 20), # A
extraDistr::rbvpois(1000, 0, 20, 20), # X
diff --git a/vignettes/getting-started-poisson-version.Rmd b/vignettes/getting-started-poisson-version.Rmd
index bc6461f46bf0edbd6855dcd2d0a77e434fa84125..7dc86b2419286b2db2c2e616500cb438fc980853 100644
--- a/vignettes/getting-started-poisson-version.Rmd
+++ b/vignettes/getting-started-poisson-version.Rmd
@@ -190,6 +190,12 @@ extraDistr::rbvpois(1000*res[[1]]$kappa, res[[1]]$l1, res[[1]]$l2, res[[1]]$l3)
# Model comparaison
+## boostrap EM sub
+
+## Bootstrap EM
+
+# Model compare
+
## XY
We simulate 3 clusters from bi-Poisson distribution
@@ -201,7 +207,7 @@ data <- rbind(
extraDistr::rbvpois(n, 0, 200, 200), # X
extraDistr::rbvpois(n, 200, 0, 5) # Y
)
-res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +
@@ -220,7 +226,7 @@ data <- rbind(
extraDistr::rbvpois(n, 100, 100, 200), # A
extraDistr::rbvpois(n, 0, 200, 200) # X
)
-res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +
@@ -238,7 +244,7 @@ We simulate 1 clusters from bi-Poisson distribution
data <- rbind(
extraDistr::rbvpois(10000, 100, 100, 200) # A
)
-res <- data %>% poiss_compare_models(nboot = 10, core = 1, max_iter = 30)
+res <- data %>% poiss_compare_models(nboot = 5, core = 1, max_iter = 30, nbatch = 20)
res %>%
pivot_longer(cols = -c("name"), names_to = "metric") %>%
ggplot(aes(x = name, y = value)) +