inst2 marginals

ref #3
OuhscBbmc · Sep 15, 2024 · 5fd93d4 · 5fd93d4
1 parent 6603a52
commit 5fd93d4
Show file tree

Hide file tree

Showing 28 changed files with 147 additions and 28 deletions.
diff --git a/analysis/institution-1/figure-png/marginals-inst2-1.png b/analysis/institution-1/figure-png/marginals-inst2-1.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-10.png b/analysis/institution-1/figure-png/marginals-inst2-10.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-11.png b/analysis/institution-1/figure-png/marginals-inst2-11.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-12.png b/analysis/institution-1/figure-png/marginals-inst2-12.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-13.png b/analysis/institution-1/figure-png/marginals-inst2-13.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-14.png b/analysis/institution-1/figure-png/marginals-inst2-14.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-15.png b/analysis/institution-1/figure-png/marginals-inst2-15.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-16.png b/analysis/institution-1/figure-png/marginals-inst2-16.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-17.png b/analysis/institution-1/figure-png/marginals-inst2-17.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-18.png b/analysis/institution-1/figure-png/marginals-inst2-18.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-19.png b/analysis/institution-1/figure-png/marginals-inst2-19.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-2.png b/analysis/institution-1/figure-png/marginals-inst2-2.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-3.png b/analysis/institution-1/figure-png/marginals-inst2-3.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-4.png b/analysis/institution-1/figure-png/marginals-inst2-4.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-5.png b/analysis/institution-1/figure-png/marginals-inst2-5.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-6.png b/analysis/institution-1/figure-png/marginals-inst2-6.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-7.png b/analysis/institution-1/figure-png/marginals-inst2-7.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-8.png b/analysis/institution-1/figure-png/marginals-inst2-8.png
diff --git a/analysis/institution-1/figure-png/marginals-inst2-9.png b/analysis/institution-1/figure-png/marginals-inst2-9.png
diff --git a/analysis/institution-1/institution-1.R b/analysis/institution-1/institution-1.R
@@ -28,7 +28,6 @@ ds <-
   # )
 
 # ---- marginals-inst1 ---------------------------------------------------------------
-# Inspect continuous variables
 TabularManifest::histogram_discrete(ds, variable_name="inst1_country_cut3")
 TabularManifest::histogram_discrete(ds, variable_name="inst1_county_usa")
 TabularManifest::histogram_discrete(ds, variable_name="inst1_status")
@@ -57,6 +56,27 @@ TabularManifest::histogram_discrete(ds, variable_name="inst1_complete")
 #   cat('TabularManifest::histogram_discrete(ds, variable_name="', column,'")\n', sep="")
 # }
 
+# ---- marginals-inst2 ---------------------------------------------------------------
+TabularManifest::histogram_discrete(  ds, "inst2_instance_count")
+TabularManifest::histogram_continuous(ds, "inst2_instance_count", bin_width = 1, rounded_digits = 1)
+TabularManifest::histogram_discrete(  ds, "inst2_client_limited")
+TabularManifest::histogram_discrete(  ds, "inst2_client_institution_single")
+TabularManifest::histogram_discrete(  ds, "inst2_client_institution_multiple")
+TabularManifest::histogram_discrete(  ds, "inst2_client_other")
+TabularManifest::histogram_discrete(  ds, "inst2_start_year")
+TabularManifest::histogram_continuous(ds, "inst2_start_year"        , bin_width = 1, rounded_digits = 1)
+TabularManifest::histogram_continuous(ds, "inst2_user_count"        , bin_width = 1000, rounded_digits = 1)
+TabularManifest::histogram_continuous(ds, "inst2_project_count"     , bin_width = 1000, rounded_digits = 1)
+TabularManifest::histogram_continuous(ds, "inst2_log_count_recent"  , bin_width = 1000000, rounded_digits = 1)
+TabularManifest::histogram_continuous(ds, "inst2_em_count"          , bin_width = 5, rounded_digits = 1)
+TabularManifest::histogram_discrete(  ds, "inst2_allow_create")
+TabularManifest::histogram_discrete(  ds, "inst2_allow_production_move")
+TabularManifest::histogram_discrete(  ds, "inst2_allow_production_change")
+TabularManifest::histogram_discrete(  ds, "inst2_allow_repeating_change")
+TabularManifest::histogram_discrete(  ds, "inst2_allow_events")
+TabularManifest::histogram_discrete(  ds, "inst2_authenticate")
+TabularManifest::histogram_discrete(  ds, "inst2_complete")
+
 # ---- scatterplots ------------------------------------------------------------
 # g1 <-
 #   ggplot(ds, aes(x=horsepower, y=quarter_mile_sec, color=forward_gear_count_f)) +

diff --git a/analysis/institution-1/institution-1.Rmd b/analysis/institution-1/institution-1.Rmd
@@ -78,12 +78,13 @@ Notes
 ---------------------------------------------------------------------------
 
 1.  The current report covers `r nrow(ds)` survey responses to the institutional survey.
-1.  Other information is available at:
+1.  Other survey information is available at:
 <https://redcap.link/bmw_institutiondata>
 
 Unanswered Questions
 ---------------------------------------------------------------------------
 
+1. Please see the full list at <https://github.com/OuhscBbmc/redcap-business-model-survey-2024/blob/main/documentation/notes-ingestion.md>
 1. What does `VS` stand for?  How was it measured?
 1. Where the cars at the Philly track measured with the same phluguerstometer and the Cleveland track?
 
@@ -95,12 +96,18 @@ Answered Questions
 Marginals
 ===========================================================================
 
-Institution 1 Instrument
+Institution 1 (Instrument)
 ---------------------------------------------------------------------------
 
 ```{r marginals-inst1, echo=echo_chunks, message=message_chunks}
 ```
 
+Institution 2 (Instrument)
+---------------------------------------------------------------------------
+
+```{r marginals-inst2, echo=echo_chunks, message=message_chunks}
+```
+
 
 Scatterplots
 ===========================================================================

diff --git a/analysis/institution-1/institution-1.html b/analysis/institution-1/institution-1.html
diff --git a/analysis/institution-1/institution-1.md b/analysis/institution-1/institution-1.md
@@ -47,12 +47,13 @@ Notes
 ---------------------------------------------------------------------------
 
 1.  The current report covers 82 survey responses to the institutional survey.
-1.  Other information is available at:
+1.  Other survey information is available at:
 <https://redcap.link/bmw_institutiondata>
 
 Unanswered Questions
 ---------------------------------------------------------------------------
 
+1. Please see the full list at <https://github.com/OuhscBbmc/redcap-business-model-survey-2024/blob/main/documentation/notes-ingestion.md>
 1. What does `VS` stand for?  How was it measured?
 1. Where the cars at the Philly track measured with the same phluguerstometer and the Cleveland track?
 
@@ -64,11 +65,16 @@ Answered Questions
 Marginals
 ===========================================================================
 
-Institution 1 Instrument
+Institution 1 (Instrument)
 ---------------------------------------------------------------------------
 
 ![](figure-png/marginals-inst1-1.png)<!-- -->![](figure-png/marginals-inst1-2.png)<!-- -->![](figure-png/marginals-inst1-3.png)<!-- -->![](figure-png/marginals-inst1-4.png)<!-- -->![](figure-png/marginals-inst1-5.png)<!-- -->![](figure-png/marginals-inst1-6.png)<!-- -->![](figure-png/marginals-inst1-7.png)<!-- -->![](figure-png/marginals-inst1-8.png)<!-- -->![](figure-png/marginals-inst1-9.png)<!-- -->![](figure-png/marginals-inst1-10.png)<!-- -->![](figure-png/marginals-inst1-11.png)<!-- -->![](figure-png/marginals-inst1-12.png)<!-- -->![](figure-png/marginals-inst1-13.png)<!-- -->![](figure-png/marginals-inst1-14.png)<!-- -->![](figure-png/marginals-inst1-15.png)<!-- -->![](figure-png/marginals-inst1-16.png)<!-- -->![](figure-png/marginals-inst1-17.png)<!-- -->![](figure-png/marginals-inst1-18.png)<!-- -->![](figure-png/marginals-inst1-19.png)<!-- -->![](figure-png/marginals-inst1-20.png)<!-- -->
 
+Institution 2 (Instrument)
+---------------------------------------------------------------------------
+
+![](figure-png/marginals-inst2-1.png)<!-- -->![](figure-png/marginals-inst2-2.png)<!-- -->![](figure-png/marginals-inst2-3.png)<!-- -->![](figure-png/marginals-inst2-4.png)<!-- -->![](figure-png/marginals-inst2-5.png)<!-- -->![](figure-png/marginals-inst2-6.png)<!-- -->![](figure-png/marginals-inst2-7.png)<!-- -->![](figure-png/marginals-inst2-8.png)<!-- -->![](figure-png/marginals-inst2-9.png)<!-- -->![](figure-png/marginals-inst2-10.png)<!-- -->![](figure-png/marginals-inst2-11.png)<!-- -->![](figure-png/marginals-inst2-12.png)<!-- -->![](figure-png/marginals-inst2-13.png)<!-- -->![](figure-png/marginals-inst2-14.png)<!-- -->![](figure-png/marginals-inst2-15.png)<!-- -->![](figure-png/marginals-inst2-16.png)<!-- -->![](figure-png/marginals-inst2-17.png)<!-- -->![](figure-png/marginals-inst2-18.png)<!-- -->![](figure-png/marginals-inst2-19.png)<!-- -->
+
 
 Scatterplots
 ===========================================================================
@@ -190,4 +196,4 @@ For the sake of documentation and reproducibility, the current report was render
 
 
 
-Report rendered by wibeasley at 2024-09-15, 13:57 -0500 in 5 seconds.
+Report rendered by wibeasley at 2024-09-15, 15:16 -0500 in 9 seconds.
diff --git a/data-public/derived/variable-label.parquet b/data-public/derived/variable-label.parquet
diff --git a/data-public/metadata/variable-label.csv b/data-public/metadata/variable-label.csv
@@ -26,3 +26,16 @@ inst1_dept_home,5,"Biostats/Analytics",5,
 inst1_dept_home,7,"Clinical Trials",6,
 inst1_dept_home,6,"Dedicated Team",7,
 inst1_dept_home,98,"Other",98,
+inst2_client,1,"limited",1,
+inst2_client,2,"institution_single",2,
+inst2_client,3,"institution_multiple",3,
+inst2_client,98,"other",98,
+inst2_allow_production_change,0,"Never",0,
+inst2_allow_production_change,1,"Yes, if no existing fields are modified",1,
+inst2_allow_production_change,2,"Yes, if empty or no existing fields are modified",2,
+inst2_allow_production_change,3,"Yes, if not critical issues exist",3,
+inst2_allow_production_change,4,"Yes, if empty or no critical issues exists",4,
+inst2_authenticate,1,"table",1,
+inst2_authenticate,2,"institution",2,
+inst2_authenticate,3,"institution & table",3,
+inst2_authenticate,98,"I don't know",98,
diff --git a/documentation/notes-ingestion.md b/documentation/notes-ingestion.md
@@ -1,5 +1,13 @@
 Notes for Ingestion Process
 ================
 
-1. For `county`, both "1" and "13" are labelled "Australia".
-1. For `program_funding`, future surveys might say "NIH CTSA/CTR Grant" instead of just "NIH CTSA Grant".
+1.  For `county`, both "1" and "13" are labelled "Australia".
+1.  For `program_funding`, future surveys might say "NIH CTSA/CTR Grant" instead of just "NIH CTSA Grant".
+1.  For `redcap_pop`, how can the checkboxes be selected for both "2" ("One Institution/Organization") and "3" ("Multiple Institutions/Organizations")?
+1.  For `redcap_instance_count`, the max is "147".  Is this likely?  If not, what is a reasonable upper bound?
+1.  If we release the (multivariate) dataset publicly, let's cap some variables so they don't identify the institution.  I think marginals are fine (eg, the earliest year was 2004), but not if it can re-identify their other responses.  Variables to cap include:
+    * `redcap_start_date`, limit lower bound to ~2010, .
+    * `active_users`, limit upper bound to ~4k.
+    * `active_projects`, limit upper bound to ~4k.
+    * `logged_events`, limit upper bound to ~1M.
+    * `em_no`, limit upper bound to ~25.
diff --git a/manipulation/institution-ellis.R b/manipulation/institution-ellis.R
@@ -180,20 +180,22 @@ ds <-
     inst1_salary_mid                                    = `sal_mid`,
     inst1_salary_senior                                 = `sal_sen`,
     inst1_complete                                      = `institutional_questionnaire_complete`,
-    # redcap_instance_count                               = `redcap_instance_count`,
-    # redcap_pop                                          = `redcap_pop`,
-    # redcap_start_date                                   = `redcap_start_date`,
-    # active_users                                        = `active_users`,
-    # active_projects                                     = `active_projects`,
-    # logged_events                                       = `logged_events`,
-    # em_no                                               = `em_no`,
-    # ccus_createprojects                                 = `ccus_createprojects`,
-    # ccus_moveprod                                       = `ccus_moveprod`,
-    # ccus_changerequests                                 = `ccus_changerequests`,
-    # ccus_repeatingsetup                                 = `ccus_repeatingsetup`,
-    # ccus_addevents                                      = `ccus_addevents`,
-    # ccus_authenticate                                   = `ccus_authenticate`,
-    # institutional_questionnaire2_complete               = `institutional_questionnaire2_complete`,
+
+    inst2_instance_count                                = `redcap_instance_count`,
+    inst2_client                                        = `redcap_pop`,
+    inst2_start_date                                    = `redcap_start_date`,
+    inst2_user_count                                    = `active_users`,
+    inst2_project_count                                 = `active_projects`,
+    inst2_log_count_recent                              = `logged_events`,
+    inst2_em_count                                      = `em_no`,
+    inst2_allow_create                                  = `ccus_createprojects`,
+    inst2_allow_production_move                         = `ccus_moveprod`,
+    inst2_allow_production_change                       = `ccus_changerequests`,
+    inst2_allow_repeating_change                        = `ccus_repeatingsetup`,
+    inst2_allow_events                                  = `ccus_addevents`,
+    inst2_authenticate                                  = `ccus_authenticate`,
+    inst2_complete                                      = `institutional_questionnaire2_complete`,
+
     # manageusers                                         = `manageusers`,
     # create                                              = `create`,
     # create_charge                                       = `create_charge`,
@@ -301,7 +303,24 @@ ds <-
     -inst1_country,
   ) #|>  View()
 
-# ds$inst1_status
+# ---- groom-institution-2 -----------------------------------------------------
+ds <-
+ds |>
+  dplyr::mutate(
+    inst2_start_year              = as.integer(lubridate::year(inst2_start_date)),
+    inst2_allow_create            = as.logical(inst2_allow_create               ),
+    inst2_allow_production_move   = as.logical(inst2_allow_production_move      ),
+    inst2_allow_repeating_change  = as.logical(inst2_allow_repeating_change     ),
+    inst2_allow_events            = as.logical(inst2_allow_events               ),
+    inst2_complete                = REDCapR::constant_to_form_completion(inst2_complete),
+  ) |>
+  map_to_checkbox("inst2_client") |>
+  map_to_radio(   "inst2_allow_production_change") |>
+  map_to_radio(   "inst2_authenticate") |>
+  # dplyr::select(tidyselect::starts_with("inst2_")) |>
+  dplyr::select(
+    -inst2_start_date
+  )
 
 # ---- reestablish-column-order ------------------------------------------------
 ds <-
@@ -310,9 +329,11 @@ ds <-
     institution_index,
     tidyselect::matches("inst1_(?!complete)", perl = TRUE), # A "negative-lookahead"
     inst1_complete,
+    tidyselect::matches("inst2_(?!complete)", perl = TRUE), # A "negative-lookahead"
+    inst2_complete,
   )
 
-# ---- verify-values -----------------------------------------------------------
+# ---- verify-values-inst1 -----------------------------------------------------------
 # OuhscMunge::verify_value_headstart(ds)
 checkmate::assert_integer(  ds$institution_index      , any.missing=F , lower=1, upper=999  , unique=T)
 checkmate::assert_character(ds$inst1_country_cut3     , any.missing=F , pattern="^.{3,9}$"  )
@@ -340,6 +361,25 @@ checkmate::assert_numeric(  ds$inst1_admin_coding_fte , any.missing=T , lower=0,
 # checkmate::assert_character(ds$inst1_salary_senior    , any.missing=T , pattern="^.{1,100}$" )
 checkmate::assert_factor(   ds$inst1_complete         , any.missing=F                       )
 
+# ---- verify-values-inst2 -----------------------------------------------------------
+checkmate::assert_integer( ds$inst2_instance_count                , any.missing=T , lower=1, upper=9999       )
+checkmate::assert_logical( ds$inst2_client_limited                , any.missing=F                             )
+checkmate::assert_logical( ds$inst2_client_institution_single     , any.missing=F                             )
+checkmate::assert_logical( ds$inst2_client_institution_multiple   , any.missing=F                             )
+checkmate::assert_logical( ds$inst2_client_other                  , any.missing=F                             )
+checkmate::assert_integer( ds$inst2_start_year                    , any.missing=T                             )
+checkmate::assert_integer( ds$inst2_user_count                    , any.missing=T , lower=20, upper=99999     )
+checkmate::assert_integer( ds$inst2_project_count                 , any.missing=T , lower=5, upper=99999      )
+checkmate::assert_numeric( ds$inst2_log_count_recent              , any.missing=T , lower=5, upper=99999999   )
+checkmate::assert_integer( ds$inst2_em_count                      , any.missing=T , lower=0, upper=999        )
+checkmate::assert_logical( ds$inst2_allow_create                  , any.missing=T                             )
+checkmate::assert_logical( ds$inst2_allow_production_move         , any.missing=T                             )
+checkmate::assert_factor(  ds$inst2_allow_production_change       , any.missing=T                             )
+checkmate::assert_logical( ds$inst2_allow_repeating_change        , any.missing=T                             )
+checkmate::assert_logical( ds$inst2_allow_events                  , any.missing=T                             )
+checkmate::assert_factor(  ds$inst2_authenticate                  , any.missing=T                             )
+checkmate::assert_factor(  ds$inst2_complete                      , any.missing=F                             )
+
 # ---- specify-columns-to-upload -----------------------------------------------
 # Print colnames that `dplyr::select()`  should contain below:
 #   cat(paste0("    ", colnames(ds), collapse=",\n"))
@@ -352,6 +392,7 @@ ds_slim <-
   # dplyr::slice(1:100) |>
   dplyr::select(
     institution_index,
+
     inst1_country_cut3,
     inst1_county_usa,
     inst1_status,
@@ -376,6 +417,24 @@ ds_slim <-
     # inst1_salary_mid,
     # inst1_salary_senior,
     inst1_complete,
+
+    inst2_instance_count,
+    inst2_client_limited,
+    inst2_client_institution_single,
+    inst2_client_institution_multiple,
+    inst2_client_other,
+    inst2_start_year,
+    inst2_user_count,
+    inst2_project_count,
+    inst2_log_count_recent,
+    inst2_em_count,
+    inst2_allow_create,
+    inst2_allow_production_move,
+    inst2_allow_production_change,
+    inst2_allow_repeating_change,
+    inst2_allow_events,
+    inst2_authenticate,
+    inst2_complete,
   )
 
 ds_slim

diff --git a/manipulation/retrieve-variable-labels.R b/manipulation/retrieve-variable-labels.R
@@ -113,6 +113,6 @@ map_to_checkbox <- function( # .variable = "inst1_funding"
   d |>
     dplyr::left_join(d_wide, by = "institution_index") |>
     dplyr::select(
-      -!!"inst1_funding"
+      -!!.variable
     )
 }