rapport-CEI-MIA-2023/Rcodes/real_data/CoOPLBM_completion_analyze.Rmd

210 lines
7.6 KiB
Text

---
title: "Netclustering analysis with the CoOPLBM completion"
bibliography: references.bib
suppress-bibliography: true
output:
html_document:
toc: true
theme: journal
pdf_document:
keep_tex: true
---
```{r libraries, echo = FALSE, include=FALSE}
devtools::load_all()
require(aricode)
```
```{r useful_functions, echo = FALSE}
extract_unlist <- function(data) {
readRDS(data) |>
extract_best_bipartite_partition() |>
unlist()
}
show_groups <- function(collection_list) {
lapply(collection_list, function(collection) {
collection$Q
})
}
partition_BICL <- function(collection_list) {
sum(sapply(collection_list, function(collection) {
collection$BICL
}))
}
extract_collection_clustering_id <- function(collection_list) {
out_dataframe <- setNames(data.frame(matrix(nrow = 0, ncol = 2)), c("netid", "collection_id"))
for (col_idx in seq_along(collection_list)) {
for (net_id in collection_list[[col_idx]]$net_id) {
out_dataframe[nrow(out_dataframe) + 1, ] <- c(net_id, col_idx)
}
}
out_dataframe
}
extract_full_models <- function(model_collections_list) {
return(do.call(
"rbind",
lapply(
seq_along(model_collections_list),
function(idx) {
model_name <- names(model_collections_list)[idx]
df <- extract_collection_clustering_id(model_collections_list[[idx]])
return(cbind(df, model = rep(model_name, length(df$netid))))
}
)
))
}
reorder_match_netids <- function(dataframe, target) {
df <- dataframe[match(target, dataframe$netid), ]
return(df)
}
extract_full_reorder <- function(model_collections_list, target) {
unordered <- extract_full_models(model_collections_list)
return(do.call("rbind", lapply(
names(model_collections_list),
function(model_name) {
return(reorder_match_netids(unordered[which(unordered$model == model_name), ], target))
}
)))
}
```
```{r data_importation, echo = FALSE}
# Uncompleted
uncompleted_model_list <- list(
"iid" = extract_unlist("real_data/data/dore_uncompleted_collection_clustering_nb_run_1_iid_70_networks_08-06-23-16:31:17.Rds"),
"pi" = extract_unlist("real_data/data/dore_uncompleted_collection_clustering_nb_run_1_pi_70_networks_08-06-23-16:52:16.Rds"),
"rho" = extract_unlist("real_data/data/dore_uncompleted_collection_clustering_nb_run_1_rho_70_networks_08-06-23-16:49:58.Rds"),
"pirho" = extract_unlist("real_data/data/dore_uncompleted_collection_clustering_nb_run_1_pirho_70_networks_08-06-23-16:41:33.Rds")
)
# Below we will need to have the netid in the same order so we choose to use the
# uncompleted iid model order as reference
netid_order <- extract_collection_clustering_id(uncompleted_model_list$iid)$netid
model_order <- c("iid", "pi", "rho", "pirho")
uncompleted_clusterings <- extract_full_reorder(uncompleted_model_list, netid_order)
# 0.2 threshold
point_2_model_list <- list(
"iid" = extract_unlist("real_data/data/dore_point_2_completed_collection_clustering_nb_run_1_iid_70_networks_07-06-23-18:40:10.Rds"),
"pi" = extract_unlist("real_data/data/dore_point_2_completed_collection_clustering_nb_run_1_pi_70_networks_07-06-23-19:22:19.Rds"),
"rho" = extract_unlist("real_data/data/dore_point_2_completed_collection_clustering_nb_run_1_rho_70_networks_07-06-23-20:03:53.Rds"),
"pirho" = extract_unlist("real_data/data/dore_point_2_completed_collection_clustering_nb_run_1_pirho_70_networks_07-06-23-21:09:12.Rds")
)
point_2_clusterings <- extract_full_reorder(point_2_model_list, netid_order)
# 0.5 threshold
point_5_model_list <- list(
"iid" = extract_unlist("real_data/data/dore_point_5_completed_collection_clustering_nb_run_1_iid_70_networks_07-06-23-19:19:53.Rds"),
"pi" = extract_unlist("real_data/data/dore_point_5_completed_collection_clustering_nb_run_1_pi_70_networks_07-06-23-21:31:20.Rds"),
"rho" = extract_unlist("real_data/data/dore_point_5_completed_collection_clustering_nb_run_1_rho_70_networks_07-06-23-21:03:50.Rds"),
"pirho" = extract_unlist("real_data/data/dore_point_5_completed_collection_clustering_nb_run_1_pirho_70_networks_07-06-23-21:13:10.Rds")
)
point_5_clusterings <- extract_full_reorder(point_5_model_list, netid_order)
# Uniform re-sampled
random_model_list <- list(
"iid" = extract_unlist("real_data/data/dore_random_completed_collection_clustering_nb_run_1_iid_70_networks_07-06-23-21:44:14.Rds"),
"pi" = extract_unlist("real_data/data/dore_random_completed_collection_clustering_nb_run_1_pi_70_networks_07-06-23-22:52:47.Rds"),
"rho" = extract_unlist("real_data/data/dore_random_completed_collection_clustering_nb_run_1_rho_70_networks_08-06-23-18:16:04.Rds"),
"pirho" = extract_unlist("real_data/data/dore_random_completed_collection_clustering_nb_run_1_pirho_70_networks_07-06-23-23:07:08.Rds")
)
random_clusterings <- extract_full_reorder(random_model_list, netid_order)
```
# Context of this analysis
After performing a netclustering on the raw data, we will see if the detect
structure resulting in the clustering comes from the sampling effort. To test
this we will use the CoOPLBM model by
@anakokDisentanglingStructureEcological2022 to complete the data.
The CoOPLBM model assumes that the observed incidence matrix $R$ is an
element-wise product of an $M$ matrix following an LBM and an $N$ matrix which
elements follow Poisson distributions independent on $M$.
The model gives us the $\widehat{M}$ matrix, the elements of which are:
$$\widehat{M_{ij}} = \mathbb{P}(M_{ij} = 1)$$
Note that if $R_{ij} = 1$ then $\widehat{M_{ij}} = 1$
- 1 if the interaction was observed
- a probability, that there should be an interaction but it wasn't observed
This *completed matrix* can be used in different manners to be fed to the colSBM
model.
# Threshold based completions
With the thresholds, the infered incidence matrix obtained by
CoOPLBM is used to generate a completed incidence matrix by the following
procedure :
$$X_{ij} = \begin{cases}
1 & \text{if the value is over the threshold} \\
0 & \text{else} \\
\end{cases}$$
## 0.5 completed threshold
```{r useful-functions, echo = FALSE, include=FALSE}
ARI_netclustering_models <- function(
clustering_compare,
uncompleted_clustering = uncompleted_clustering,
models = c("iid", "pi", "rho", "pirho")) {
sapply(models, function(model) {
ARI(
uncompleted_clusterings[
which(uncompleted_clusterings$model == model),
]$collection_id,
clustering_compare[
which(clustering_compare$model == model),
]$collection_id
)
})
}
```
Here, the completion threshold is set to $0.5$.
First we will compute an ARI on the collection id given by the raw data and the
completed matrix.
```{r 0.5_ARI, echo = FALSE, results="asis"}
knitr::kable(ARI_netclustering_models(point_5_clusterings),
col.names = c("ARI with uncompleted data")
)
```
In the above table, one can see the network clustering obtained after applying
CoOPLBM has not much in common with the clustering of the uncompleted data.
### Number of sub-collections and details of each sub-collection
```{r 0.5_partition_numbers, echo = FALSE}
```
## 0.2 completed threshold
The $0.2$ threshold adds a lot of interactions compared to raw matrix.
```{r 0.2_ARI, echo = FALSE, results="asis"}
knitr::kable(ARI_netclustering_models(point_2_clusterings),
col.names = c("ARI with uncompleted data")
)
```
# Sample based completions
The $M$ matrix is used to sample a new $X$ matrix which elements are the
realisation of Bernoulli distributions of probability $M_{i,j}$.
$$\mathbb{P}(X_{i,j} = 1) = M_{i,j} $$
```{r random_ARI, echo = FALSE, results="asis"}
knitr::kable(ARI_netclustering_models(random_clusterings),
col.names = c("ARI with uncompleted data")
)
```