full synthetic workflow

various changes

.Renviron

@@ -0,0 +1,2 @@
+PYTHON_BIN="/Users/bertrandgallice/opt/miniconda3/envs/mlflow_env/bin/python"
+PYTHON_MLFLOW_BIN=""/Users/bertrandgallice/opt/miniconda3/envs/mlflow_env/bin/mlflow""

.gitignore

@@ -8,9 +8,8 @@
 drafts/
 legacy/
 documentation/
-data-raw/
 inst/doc
 /doc/
 /Meta/
 workspace/
-
+data-raw/mlflow_outputs/

DESCRIPTION

@@ -11,6 +11,9 @@ Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.2.3
 Imports: 
+    2DegreesInvesting/r2dii.climate.stress.test,
+    2DegreesInvesting/STDataMGMT,
+    box,
     dplyr,
     forcats,
     ggplot2,

R/data_load_crispy.R

@@ -1,11 +1,4 @@
-#' Title
-#'
-#' @param crispy_outputs_dir
-#'
-#' @return
-#' @export
-#'
-#' @examples
+
 load_multiple_crispy <- function(crispy_outputs_dir, max_granularity) {
   # Required Libraries
 
@@ -56,16 +49,13 @@ load_multiple_crispy <- function(crispy_outputs_dir, max_granularity) {
 #'
 #' # TODO FIND CLOSEST COMPANY IF group_cols=NULL
 #'
-#' @param multi_crispy
-#' @param group_cols
+#' @param multi_crispy multi_crispy
+#' @param group_cols group_cols
 #' @param param_cols TODO the parameters should be stored in a dataframe other than crispy and use the run_id as key to join
 #'
-#' @return
-#'
-#' @examples
 aggregate_crispy_facts <- function(multi_crispy, group_cols, param_cols = c(
                                      "run_id", "roll_up_type","scenario_geography","baseline_scenario",
-                                     "shock_scenario","lgd","risk_free_rate","discount_rate","div_netprofit_prop_coef",
+                                     "shock_scenario","risk_free_rate","discount_rate","div_netprofit_prop_coef",
                                      "carbon_price_model","market_passthrough","financial_stimulus","start_year",
                                      "growth_rate","shock_year")
                                      ) {

R/data_load_portfolio.R

@@ -1,11 +1,3 @@
-#' Title
-#'
-#' @param portfolio_data_path
-#'
-#' @return
-#' @export
-#'
-#' @examples
 load_portfolio_data <- function(portfolio_data_path) {
   if (!is.null(portfolio_data_path)) {
     portfolio_data <- readr::read_csv(
@@ -42,16 +34,12 @@ load_portfolio_data <- function(portfolio_data_path) {
 }
 
 
-#' Title
+#' Convert column year to term
 #'
-#' @param portfolio_data
-#' @param maturity_month_term_bridge
-#' @param start_year
+#' @param portfolio_data portfolio_data
+#' @param maturity_month_term_bridge maturity_month_term_bridge
+#' @param start_year start_year
 #'
-#' @return
-#' @export
-#'
-#' @examples
 map_portfolio_maturity_to_term <-
   function(portfolio_data,
            maturity_month_term_bridge,

R/data_preproc_analysis.R

@@ -1,11 +1,4 @@
-#' Title
-#'
-#' @param portfolio_data
-#' @param multi_crispy
-#'
-#' @return
-#'
-#' @examples
+
 create_analysis_data <-
   function(portfolio_data,
            multi_crispy_data,
@@ -51,17 +44,17 @@ compute_analysis_metrics <- function(analysis_data) {
     dplyr::mutate(
       net_present_value_difference = .data$crispy.net_present_value_shock - .data$crispy.net_present_value_baseline,
       crispy_perc_value_change = .data$net_present_value_difference / .data$crispy.net_present_value_baseline,
-      crispy_value_loss = .data$crispy_perc_value_change * .data$portfolio.exposure_value_usd,
+      crispy_value_loss = .data$crispy_perc_value_change * .data$exposure_value_usd,
 
-      exposure_at_default = .data$portfolio.exposure_value_usd * .data$portfolio.loss_given_default,
+      exposure_at_default = .data$exposure_value_usd * .data$loss_given_default,
       # exposure_at_default_baseline = .data$net_present_value_baseline * .data$loss_given_default,
       # exposure_at_default_shock = .data$net_present_value_shock * .data$loss_given_default,
 
-      expected_loss_portfolio = .data$exposure_at_default * .data$portfolio.pd_portfolio,
+      expected_loss_portfolio = .data$exposure_at_default * .data$pd_portfolio,
       expected_loss_baseline = .data$exposure_at_default * .data$crispy.pd_baseline,
       expected_loss_shock = .data$exposure_at_default * .data$crispy.pd_shock,
 
-      # pd_difference_portfolio = .data$portfolio.pd_portfolio - .data$crispy.pd_shock,
+      # pd_difference_portfolio = .data$pd_portfolio - .data$crispy.pd_shock,
       pd_difference = .data$crispy.pd_shock - .data$crispy.pd_baseline,
       crispy_perc_pd_change = .data$pd_difference / .data$crispy.pd_baseline
     )

R/draw_cdi_plots.R

@@ -78,7 +78,7 @@ make_mean_pd_diff_plot <- function(data_cdi_pd_plot, scenario_name_for_title) {
       labels = scales::comma,
       name = "Mean pd_difference"
     ) +
-    facet_grid( ~ portfolio.ald_sector, scales = "free_y") + # Allow separate scales for y-axis
+    facet_grid( ~ ald_sector, scales = "free_y") + # Allow separate scales for y-axis
     theme_2dii() +
     labs(x = "Shock_Year", y = "Mean pd_difference", title = paste("Mean PD Difference by Shock Year -", scenario_name_for_title))
   # theme(plot.title = element_text(hjust = 0.5, face = "bold", size = 28),  # Adjust title size
@@ -116,7 +116,7 @@ make_discount_rate_plot <- function(plot_data, scenario_name_for_title, y_var) {
       labels = scales::comma,
       name = "Mean pd_difference"
     ) +
-    facet_grid( ~ portfolio.ald_sector, scales = "free_y") + # Allow separate scales for y-axis
+    facet_grid( ~ ald_sector, scales = "free_y") + # Allow separate scales for y-axis
     theme_2dii() +
     labs(x = "Discount Rate", y = "Mean pd_difference", title = paste("Mean PD Difference by Discount Rate -", scenario_name_for_title))
   # theme(plot.title = element_text(hjust = 0.5, face = "bold", size = 24),  # Adjust title size
@@ -142,7 +142,7 @@ make_discount_rate_plot <- function(plot_data, scenario_name_for_title, y_var) {
 #' @export
 #'
 #' @examples
-make_density_plots <- function(data_cdi_pd_plot, numeric_values, density_var, group_variable="portfolio.ald_sector") {
+make_density_plots <- function(data_cdi_pd_plot, numeric_values, density_var, group_variable="ald_sector") {
 
   density_var_values <- unique(data_cdi_pd_plot[[density_var]])
   plots <- list()

R/main_data_load.R

@@ -4,34 +4,81 @@
 #'  The dataframe in output of this function should always be
 #'  the one used as input for the plots preprocessing functions
 #'
-#' @param crispy_outputs_dir
-#' @param portfolio_data_path
-#' @param granularity
-#' @param maturity_month_term_bridge_fp
+#' @param crispy_outputs_dir crispy_outputs_dir
+#' @param portfolio_data_path portfolio_data_path
+#' @param granularity granularity
+#' @param maturity_month_term_bridge_fp see in data-raw in the source package
+#' @param trisk_start_year (default) sets to the earliest year of multi_cripy_data
 #'
 #' @return
 #' @export
 #'
-#' @examples
-load_input_plots_data <-
+load_input_plots_data_from_files <-
   function(crispy_outputs_dir,
            portfolio_data_path = NULL,
            granularity = c("company_id", "company_name", "ald_sector", "ald_business_unit"),
            maturity_month_term_bridge_fp = here::here("data", "maturity_month_term_bridge.csv"),
-           trisk_start_year = 2022) {
+           trisk_start_year = NA) {
+
+    multi_crispy_data <-
+      load_multiple_crispy(crispy_outputs_dir = crispy_outputs_dir) |>
+      main_load_multi_crispy_data(
+        max_crispy_granularity =
+          c("run_id", "term", "scenario_geography", granularity)
+      )
+
+    portfolio_data <- load_portfolio_data(portfolio_data_path) |>
+      main_load_portfolio_data(
+        maturity_month_term_bridge_fp = maturity_month_term_bridge_fp,
+        max_portfolio_granularity = c("portfolio_id", "term", granularity),
+        trisk_start_year = min(multi_crispy_data$year)
+      )
+
+    analysis_data <-
+      main_load_analysis_data(
+        portfolio_data = portfolio_data,
+        multi_crispy_data = multi_crispy_data,
+        portfolio_crispy_merge_cols = c("term", granularity)
+      )
+
+    return(analysis_data)
+  }
+
+#' Data load function to generate plots
+#'
+#' @description
+#'  The dataframe in output of this function should always be
+#'  the one used as input for the plots preprocessing functions
+#'
+#' @param crispy_outputs_dir crispy_outputs_dir
+#' @param portfolio_data_path portfolio_data_path
+#' @param granularity granularity
+#' @param maturity_month_term_bridge_fp see in data-raw in the source package
+#' @param trisk_start_year (default) sets to the earliest year of multi_cripy_data
+#'
+#' @return
+#' @export
+#'
+load_input_plots_data_from_tibble <-
+  function(multi_crispy_data,
+           portfolio_data = tibble::tibble(),
+           granularity = c("company_id", "company_name", "ald_sector", "ald_business_unit"),
+           maturity_month_term_bridge_fp = here::here("data", "maturity_month_term_bridge.csv"),
+           trisk_start_year = NA) {
+
     multi_crispy_data <-
+      multi_crispy_data |>
       main_load_multi_crispy_data(
-        crispy_outputs_dir = crispy_outputs_dir,
         max_crispy_granularity =
           c("run_id", "term", "scenario_geography", granularity)
       )
 
     portfolio_data <-
+      portfolio_data |>
       main_load_portfolio_data(
-        portfolio_data_path = portfolio_data_path,
         maturity_month_term_bridge_fp = maturity_month_term_bridge_fp,
         max_portfolio_granularity = c("portfolio_id", "term", granularity),
-        trisk_start_year = trisk_start_year
+        trisk_start_year = min(multi_crispy_data$year)
       )
 
     analysis_data <-
@@ -44,17 +91,18 @@ load_input_plots_data <-
     return(analysis_data)
   }
 
+
 #' Title
 #'
 #'
 #'
-#' @param multi_crispy
-#' @param portfolio_data
-#' @param portfolio_crispy_merge_cols
+#' @param multi_crispy multi_crispy
+#' @param portfolio_data portfolio_data
+#' @param portfolio_crispy_merge_cols portfolio_crispy_merge_cols
 #'
 #' @return
 #'
-#' @examples
+#' @export
 main_load_analysis_data <-
   function(multi_crispy_data,
            portfolio_data,
@@ -71,17 +119,16 @@ main_load_analysis_data <-
 
 #' Title
 #'
-#' @param max_portfolio_granularity
+#' @param max_portfolio_granularity max_portfolio_granularity
 #'
 #' @return
 #'
-#' @examples
+#' @export
 main_load_portfolio_data <-
-  function(portfolio_data_path,
+  function(portfolio_data,
            maturity_month_term_bridge_fp,
            max_portfolio_granularity,
            trisk_start_year) {
-    portfolio_data <- load_portfolio_data(portfolio_data_path)
     maturity_month_term_bridge <- readr::read_csv(maturity_month_term_bridge_fp)
 
     portfolio_data <- portfolio_data |>
@@ -96,17 +143,16 @@ main_load_portfolio_data <-
 
 #' Title
 #'
-#' @param crispy_outputs_dir
-#' @param max_crispy_granularity
+#' @param crispy_outputs_dir crispy_outputs_dir
+#' @param max_crispy_granularity max_crispy_granularity
 #'
 #' @return
 #'
-#' @examples
+#' @export
 main_load_multi_crispy_data <-
-  function(crispy_outputs_dir,
+  function(multi_crispy_data,
            max_crispy_granularity) {
-    multi_crispy_data <-
-      load_multiple_crispy(crispy_outputs_dir = crispy_outputs_dir) |>
+    multi_crispy_data <- multi_crispy_data |>
       aggregate_crispy_facts(group_cols = max_crispy_granularity)
     return(multi_crispy_data)
   }