Merge pull request #27 from egouldo/summary-tbl-funs

Create functions implementing summary tables

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: ManyEcoEvo
 Title: Meta-analyse data from 'Many-Analysts' style studies
-Version: 1.0.0
+Version: 1.0.0.9000
 Authors@R: c(person(given = "Elliot",
                     family = "Gould", 
                     email =  "[email protected]", 

NAMESPACE

@@ -22,6 +22,9 @@ export(compute_metaanalysis_inputs)
 export(conversion)
 export(conversion_2)
 export(convert_predictions)
+export(count_analyses_variables_used)
+export(count_binary_coded_features)
+export(count_teams_analyses)
 export(cube_back)
 export(divide_back)
 export(est_to_zr)
@@ -72,17 +75,48 @@ export(square_root_back)
 export(standardise_response)
 export(subset_fns_Zr)
 export(subset_fns_yi)
+export(summarise_analyses_by_reviewer)
+export(summarise_analysis_types)
+export(summarise_conclusions)
+export(summarise_conclusions_data)
+export(summarise_model_composition)
+export(summarise_model_composition_data)
+export(summarise_reviews)
+export(summarise_reviews_per_analysis)
+export(summarise_sorensen_index)
+export(summarise_sorensen_index_data)
+export(summarise_variable_counts)
 export(validate_predictions)
 export(validate_predictions_df_blue_tit)
 export(validate_predictions_df_euc)
+import(metafor)
+importFrom(broom,tidy)
+importFrom(dplyr,across)
 importFrom(dplyr,case_when)
+importFrom(dplyr,count)
+importFrom(dplyr,distinct)
+importFrom(dplyr,ends_with)
+importFrom(dplyr,everything)
+importFrom(dplyr,filter)
+importFrom(dplyr,full_join)
+importFrom(dplyr,group_by)
 importFrom(dplyr,group_split)
 importFrom(dplyr,left_join)
+importFrom(dplyr,mutate)
+importFrom(dplyr,n_distinct)
 importFrom(dplyr,rename)
+importFrom(dplyr,right_join)
 importFrom(dplyr,select)
+importFrom(dplyr,summarise)
 importFrom(magrittr,"%>%")
 importFrom(pointblank,col_vals_not_null)
+importFrom(purrr,map)
+importFrom(purrr,map_dfr)
+importFrom(purrr,set_names)
 importFrom(rlang,is_na)
 importFrom(rlang,na_chr)
+importFrom(tidyr,pivot_longer)
+importFrom(tidyr,pivot_wider)
 importFrom(tidyr,separate)
 importFrom(tidyr,unite)
+importFrom(tidyr,unnest)

R/generate_yi_subsets.R

@@ -6,6 +6,7 @@
 #' @family Multi-dataset Wrapper Functions
 #' @return A `datraframe` with the character columns `dataset`, `estimate_type` and list-cols `data` and `diversity_data`
 #' @export
+#' @author Elliot Gould
 generate_yi_subsets <- function(yi_analysis){
 
   # So: yi_analysis$data[[1]]$back_transformed data is that obj.

R/summarise_analysis_types.R

@@ -0,0 +1,161 @@
+#' Summarise Analysis Types
+#' @description
+#' Generates a summary of the number of analysis teams, total analyses, models with normal error distributions, mixed effects models, and models developed using Bayesian statistical methods for a given analysis type.
+#' 
+#' @details
+#' Applies both [count_binary_coded_features()] and [count_teams_analyses()] to generate data summaries.
+#' 
+#' 
+#' @param ManyEcoEvo_results A tibble of `ManyEcoEvo_results`
+#' @param ManyEcoEvo_yi_results A tibble of `ManyEcoEvo_yi_results`
+#' @param ManyEcoEvo A tibble of `ManyEcoEvo`
+#'
+#' @return A summarised tibble with the variables `subset`, `dataset`, `num_teams`, `total_analyses`, `sum_linear`, `sum_mixed`, `sum_Bayesian`.
+#' @export
+#' @importFrom dplyr filter
+#' @importFrom dplyr select
+#' @importFrom dplyr group_by
+#' @importFrom dplyr rename
+#' @importFrom purrr map
+#' @importFrom broom tidy
+#' @importFrom tidyr unnest
+#' @importFrom magrittr "%>%"
+#' @importFrom dplyr distinct
+#' @importFrom dplyr mutate
+#' @import metafor
+#' @importFrom dplyr across
+#' @importFrom dplyr left_join
+#' @importFrom dplyr right_join
+#' @importFrom dplyr full_join
+#' @importFrom purrr map_dfr
+#' @family Multi-dataset Wrapper Functions
+#' @author Hannah S. Fraser
+#' @author Elliot Gould
+#' @examples
+#' summarise_analysis_types(ManyEcoEvo_results, ManyEcoEvo_yi_results, ManyEcoEvo)
+summarise_analysis_types <- function(ManyEcoEvo_results, ManyEcoEvo_yi_results, ManyEcoEvo) {
+  effect_ids <- ManyEcoEvo_results %>% 
+    filter(exclusion_set == "complete", 
+           publishable_subset == "All") %>% 
+    select(MA_mod, effects_analysis) %>% 
+    group_by(estimate_type, dataset) %>% 
+    mutate(tidy_mod = map(MA_mod, 
+                             ~ broom::tidy(.x, 
+                                           conf.int = TRUE, 
+                                           include_studies = TRUE) %>% 
+                               rename(study_id = term)), .keep = "none") %>% 
+    unnest(tidy_mod) %>% 
+    filter(type =="study") %>% 
+    ungroup %>% 
+    select(study_id) %>% 
+    rename(id_col = study_id) %>% #TODO duplicates for "Bell-2-2-1" and "Bonalbo-1-1-1 WHY?
+    distinct()
+
+  prediction_ids <- ManyEcoEvo_yi_results %>% #TODO Euc mod_data_logged not here!
+    filter(exclusion_set == "complete", 
+           # dataset == "blue tit"
+           ) %>% 
+    select(MA_mod, effects_analysis, -exclusion_set) %>% 
+    group_by(estimate_type, dataset) %>% 
+    mutate(tidy_mod = map(MA_mod, 
+                             ~ broom::tidy(.x, conf.int = TRUE, include_studies = TRUE) %>% 
+                               rename(study_id = term)), .keep = "none") %>% 
+    unnest(tidy_mod) %>% filter(type == "study") %>% 
+    ungroup %>% 
+    select(study_id) %>% 
+    rename(id_col = study_id) %>% 
+    distinct()
+
+  Master <-
+    ManyEcoEvo %>% 
+    ungroup %>% 
+    select(data) %>% 
+    unnest(data) %>% 
+    mutate(across(c(num_fixed_variables,
+                    num_random_variables,
+                    sample_size,
+                    num_interactions,
+                    Bayesian,
+                    mixed_model), 
+                  as.numeric),
+           lm = ifelse(linear_model == "linear", 1, 0), 
+           glm = ifelse(linear_model == "generalised", 1, 0)) #TODO move this into master processing so don't have to repeat else where!!
+
+  effects <- Master %>% 
+    right_join(effect_ids, by = c("id_col")) # repeat for each
+
+  predictions <- Master %>% 
+    right_join(prediction_ids, by = c("id_col"))
+
+  summarised_data <- full_join(
+  map_dfr(.x = list(effects, predictions) %>% 
+            purrr::set_names("effects", "predictions"),
+          count_teams_analyses, 
+          .id = "subset"),
+      map_dfr(.x = list(effects, predictions) %>% 
+            purrr::set_names("effects", "predictions"),
+          count_binary_coded_features,
+          .id = "subset")
+  )
+
+  return(summarised_data)
+  #TODO next: set up so can run on just one object ManyEcoEvo_results, and account for subsets too!
+}
+
+
+#' Summarise number of analyst teams and total analyses per dataset
+#'
+#' @param data A dataframe containing the variables `TeamIdentifier` and `dataset`
+#'
+#' @return A dataframe with the columns  `dataset`, `total_teams` and `total_analyses` equal in number of rows to the number of unique values within the `dataset` variable of the input `data`.
+#' @export
+#' @importFrom dplyr count
+#' @importFrom dplyr group_by
+#' @importFrom dplyr summarise
+#' @importFrom magrittr "%>%"
+#' @author Hannah S. Fraser
+#' @author Elliot Gould
+#' @examples
+#' ManyEcoEvo %>% 
+#' filter(dataset == "blue tit") %>% 
+#' ungroup %>% 
+#' select(data) %>% 
+#' unnest(data) %>% 
+#' count_teams_analyses()
+count_teams_analyses <- function(data) {
+  data %>% 
+  count(dataset, TeamIdentifier) %>% #TODO consider renaming col
+    group_by(dataset) %>% 
+    summarise(total_teams = n(), 
+              total_analyses = sum(n))
+}
+
+#' Summarise binary coded features of analyses
+#'
+#' @param data A dataframe containing the variables `dataset`, and `lm`, `mixed_model`, `Bayesian`, which are coded as binary numeric vectors.
+#'
+#' @return A dataframe with the variables
+#' @export
+#' @importFrom dplyr group_by
+#' @importFrom dplyr summarise
+#' @importFrom magrittr "%>%"
+#' @author Hannah S. Fraser
+#' @author Elliot Gould
+#' @examples
+#' ManyEcoEvo %>%
+#' filter(dataset == "eucalyptus") %>%
+#' ungroup %>%
+#'  select(data) %>%
+#'  unnest(data) %>%
+#'  mutate(lm = ifelse(linear_model == "linear", 1, 0), #TODO move into master processing
+#'         glm = ifelse(linear_model == "generalised", 1, 0),
+#'         Bayesian = as.numeric(Bayesian)) %>% 
+#'  count_binary_coded_features()
+count_binary_coded_features <- function(data){
+  data %>% 
+    group_by(dataset) %>%  
+    summarise(sum_linear = sum(lm, na.rm = TRUE), 
+              sum_mixed = sum(mixed_model,na.rm = TRUE),
+              sum_Bayesian = sum(Bayesian,na.rm = TRUE),
+              sum_glm = sum(glm, na.rm = TRUE))
+}

R/summarise_conclusions.R

@@ -0,0 +1,129 @@
+#' Summarise counts of qualitative conclusions across all datasets
+#' 
+#' @details
+#' Data summary is generated with [summarise_conclusions_data()].
+#' 
+#' @param ManyEcoEvo_results A tibble of `ManyEcoEvo_results`
+#' @param ManyEcoEvo_yi_results A tibble of `ManyEcoEvo_yi_results`
+#' @param ManyEcoEvo A tibble of `ManyEcoEvo`
+#'
+#' @return A dataframe with count values for each unique `Conclusion` in columns for each `subset` ("effects", "predictions", "all"), for each `dataset`.
+#' @export
+#' @importFrom dplyr filter
+#' @importFrom dplyr select
+#' @importFrom dplyr group_by
+#' @importFrom dplyr rename
+#' @importFrom purrr map
+#' @importFrom broom tidy
+#' @importFrom tidyr unnest
+#' @importFrom magrittr "%>%"
+#' @importFrom dplyr distinct
+#' @importFrom dplyr mutate
+#' @import metafor
+#' @importFrom dplyr across
+#' @importFrom dplyr left_join
+#' @importFrom dplyr right_join
+#' @importFrom dplyr full_join
+#' @importFrom purrr map_dfr
+#' @importFrom tidyr pivot_wider
+#' @family Multi-dataset Wrapper Functions
+#' @author Hannah S. Fraser
+#' @author Elliot Gould
+#' @examples
+#' summarise_conclusions(ManyEcoEvo_results,ManyEcoEvo_yi_results,ManyEcoEvo)
+summarise_conclusions <- function(ManyEcoEvo_results, ManyEcoEvo_yi_results, ManyEcoEvo) {
+  effect_ids <- ManyEcoEvo_results %>% 
+    filter(exclusion_set == "complete", 
+           publishable_subset == "All") %>% 
+    ungroup %>% 
+    select(MA_mod, effects_analysis, estimate_type, dataset) %>% 
+    group_by(estimate_type, dataset) %>% 
+    mutate(tidy_mod = map(MA_mod, 
+                          ~ broom::tidy(.x, 
+                                        conf.int = TRUE, 
+                                        include_studies = TRUE) %>% 
+                            rename(study_id = term)), .keep = "none") %>% 
+    unnest(tidy_mod) %>% 
+    filter(type =="study") %>% 
+    ungroup %>% 
+    select(study_id) %>% 
+    rename(id_col = study_id) %>% #TODO duplicates for "Bell-2-2-1" and "Bonalbo-1-1-1 WHY?
+    distinct()
+
+  prediction_ids <- ManyEcoEvo_yi_results %>% #TODO Euc mod_data_logged not here!
+    filter(exclusion_set == "complete", 
+           # dataset == "blue tit"
+    ) %>% 
+    ungroup %>% 
+    select(MA_mod, effects_analysis, -exclusion_set, dataset, estimate_type) %>% 
+    group_by(estimate_type, dataset) %>% 
+    mutate(tidy_mod = map(MA_mod, 
+                          ~ broom::tidy(.x, conf.int = TRUE, include_studies = TRUE) %>% 
+                            rename(study_id = term)), .keep = "none") %>% 
+    unnest(tidy_mod) %>% filter(type == "study") %>% 
+    ungroup %>% 
+    select(study_id) %>% 
+    rename(id_col = study_id) %>% 
+    distinct()
+
+  Master <-
+    ManyEcoEvo %>% 
+    ungroup %>% 
+    select(data) %>% 
+    unnest(data) %>% 
+    mutate(across(c(num_fixed_variables,
+                    num_random_variables,
+                    sample_size,
+                    num_interactions,
+                    Bayesian,
+                    mixed_model), 
+                  as.numeric),
+           lm = ifelse(linear_model == "linear", 1, 0), 
+           glm = ifelse(linear_model == "generalised", 1, 0)) %>%  #TODO move this into master processing so don't have to repeat else where!!
+    filter(Conclusion != "CHECK", !is.na(Conclusion)) #TODO data cleaning, check these
+
+  effects <- Master %>% 
+    right_join(effect_ids, by = c("id_col")) # repeat for each
+
+  predictions <- Master %>% 
+    right_join(prediction_ids, by = c("id_col"))
+
+  summarised_data <- 
+    map_dfr(.x = list(effects, predictions, Master) %>% 
+              purrr::set_names("effects", "predictions", "all"),
+            .f = summarise_conclusions_data,
+            .id = "subset") %>% 
+    pivot_wider(names_from = Conclusion, 
+                values_from = n,
+                values_fill = 0) %>% 
+    ungroup()
+
+  return(summarised_data)
+
+}
+
+
+#' Count qualitative conclusions across all analyses for each dataset
+#'
+#' @param data A dataframe containing the columns `split_id`, `analysis_id`, `dataset`, `Conclusion`
+#'
+#' @return A dataframe with counts `n` for each unique value of `Conclusion` for each `dataset`
+#' @export
+#' @importFrom dplyr filter
+#' @importFrom dplyr group_by
+#' @importFrom dplyr count
+#' @importFrom magrittr "%>%"
+#' @author Hannah S. Fraser
+#' @author Elliot Gould
+#' @examples
+#' ManyEcoEvo$data[[1]] %>%
+#' filter(Conclusion != "CHECK") %>%
+#' summarise_conclusions_data()
+summarise_conclusions_data <- function(data){
+  data %>% 
+    ungroup %>% 
+    filter(split_id == "1", analysis_id == "1" ) %>% #TODO how to generalise to data without split_id
+    group_by(dataset, Conclusion) %>% 
+    count() %>% 
+    ungroup()
+}