shuffle_spec refactoring

yjunechoe · Jul 25, 2023 · 667b9bd · 667b9bd
1 parent a1672e7
commit 667b9bd
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 24 deletions.
diff --git a/R/permute.R b/R/permute.R
@@ -68,7 +68,7 @@ permute_by_predictor <- function(jlmer_spec, predictors, predictor_type = c("gue
   time <- jlmer_spec$meta$time
   predictor_type <- match.arg(predictor_type)
   if (predictor_type == "guess") {
-    predictor_type <- .jlmerclusterperm$jl$guess_shuffle_as(df_jl, predictors, subject, trial)
+    predictor_type <- .jlmerclusterperm$jl$guess_shuffle_spec(df_jl, predictors, subject, trial)$shuffle_type
     cli::cli_alert_info("Guessed {.arg predictor_type} to be {.val {predictor_type}}")
   }
   predictor_group <- Filter(function(x) any(predictors %in% x), jlmer_spec$meta$term_groups)

diff --git a/docs/articles/Garrison-et-al-2020.html b/docs/articles/Garrison-et-al-2020.html
diff --git a/inst/julia/JlmerClusterPerm/src/05-permute.jl b/inst/julia/JlmerClusterPerm/src/05-permute.jl
@@ -1,14 +1,4 @@
-function guess_and_shuffle_as!(
-    df::DataFrame,
-    predictor_cols::Union{String,Vector{String}},
-    participant_col::String,
-    trial_col::Union{Nothing,Integer,String},
-)
-    shuffle_type = guess_shuffle_as(df, predictor_cols, participant_col, trial_col)
-    shuffle_as!(df, predictor_cols, participant_col, trial_col, shuffle_type)
-end
-
-function guess_shuffle_as(
+function guess_shuffle_spec(
     df::DataFrame,
     predictor_cols::Union{String,Vector{String}},
     participant_col::String,
@@ -17,30 +7,45 @@ function guess_shuffle_as(
     subj_pred_pair = unique(df[!, vcat(participant_col, predictor_cols)])
     unique_combinations = length(unique(df[!, participant_col])) == nrow(subj_pred_pair)
     if unique_combinations
-        "between_participant"
+        shuffle_type = "between_participant"
     else
         if (ismissing(trial_col) || isnothing(trial_col))
             throw("""Guessed "within_participant" but no column for `trial` supplied.""")
         else
-            "within_participant"
+            shuffle_type = "within_participant"
         end
     end
+    make_shuffle_spec(df, shuffle_type, predictor_cols, participant_col)
 end
 
-function shuffle_as!(
+function make_shuffle_spec(
     df::DataFrame,
     shuffle_type::String,
     predictor_cols::Union{String,Vector{String}},
     participant_col::String,
+)
+    if shuffle_type == "between_participant"
+      subj_pred_pair = unique(df[!, vcat(participant_col, predictor_cols)])
+      (shuffle_type = shuffle_type, subj_pred_pair = subj_pred_pair)
+    elseif shuffle_type == "within_participant"
+      (shuffle_type = shuffle_type,)
+    end
+end
+
+function shuffle_as!(
+    df::DataFrame,
+    shuffle_spec::NamedTuple,
+    predictor_cols::Union{String,Vector{String}},
+    participant_col::String,
     trial_col::Union{Nothing,Integer,String},
     rng::AbstractRNG,
 )
-    if shuffle_type == "between_participant"
-        subj_pred_pair = unique(df[!, vcat(participant_col, predictor_cols)])
+    if shuffle_spec.shuffle_type == "between_participant"
+        subj_pred_pair = shuffle_spec.subj_pred_pair
         shuffle!(rng, subj_pred_pair[!, participant_col])
         select!(df, Not(predictor_cols))
         leftjoin!(df, subj_pred_pair, on = participant_col)
-    elseif shuffle_type == "within_participant"
+    elseif shuffle_spec.shuffle_type == "within_participant"
         trial_pred_pair = unique(df[!, vcat(participant_col, trial_col, predictor_cols)])
         combine(
             groupby(trial_pred_pair, participant_col),
@@ -61,10 +66,11 @@ function permute_by_predictor(
     global_opts::NamedTuple,
 )
     _df = copy(df)
+    shuffle_spec = make_shuffle_spec(df, shuffle_type, predictor_cols, participant_col)
     out = insertcols(
         shuffle_as!(
             _df,
-            shuffle_type,
+            shuffle_spec,
             predictor_cols,
             participant_col,
             trial_col,
@@ -79,7 +85,7 @@ function permute_by_predictor(
                 insertcols(
                     shuffle_as!(
                         _df,
-                        shuffle_type,
+                        shuffle_spec,
                         predictor_cols,
                         participant_col,
                         trial_col,

diff --git a/inst/julia/JlmerClusterPerm/src/06-permute_timewise_statistics.jl b/inst/julia/JlmerClusterPerm/src/06-permute_timewise_statistics.jl
@@ -53,7 +53,7 @@ function permute_timewise_statistics(
     for term_groups in term_groups_est
         predictors = term_groups.p
         permute_data = copy(data)
-        shuffle_type = guess_shuffle_as(
+        shuffle_spec = guess_shuffle_spec(
             permute_data,
             predictors,
             participant_col,
@@ -70,7 +70,7 @@ function permute_timewise_statistics(
         for i = 1:nsim
             shuffle_as!(
                 permute_data,
-                shuffle_type,
+                shuffle_spec,
                 predictors,
                 participant_col,
                 trial_col,