Creating separate distributions for activity creation per group; #4

matsim-melbourne · Mar 12, 2021 · 761d7a1 · 761d7a1
1 parent 0032f18
commit 761d7a1
Show file tree

Hide file tree

Showing 27 changed files with 2,212 additions and 20 deletions.
diff --git a/R/setup.R b/R/setup.R
@@ -1,11 +1,32 @@
-# Function to pre-process some data; need only be run once
+demand_setup_groups<-function(groups,
+                              setupDir,
+                              vista18TripsCsv,
+                              out_weekday_activities_csv_gz_prefix,
+                              out_weekend_activities_csv_gz_prefix,
+                              out_weekday_activities_time_bins_csv_gz_prefix,
+                              out_weekend_activities_time_bins_csv_gz_prefix,
+                              out_weekday_activities_end_time_dist_by_start_bins_csv_gz_prefix
+                              ) {
+  for (gid in groups) {
+    demand_setup(
+      setupDir,
+      vista18TripsCsv,
+      paste0(setupDir,"/",out_weekday_activities_csv_gz_prefix,gid,".csv.gz"),
+      paste0(setupDir,"/",out_weekend_activities_csv_gz_prefix,gid,".csv.gz"),
+      paste0(setupDir,"/",out_weekday_activities_time_bins_csv_gz_prefix,gid,".csv.gz"),
+      paste0(setupDir,"/",out_weekend_activities_time_bins_csv_gz_prefix,gid,".csv.gz"),
+      paste0(setupDir,"/",out_weekday_activities_end_time_dist_by_start_bins_csv_gz_prefix,gid,".csv.gz")
+    )
+  }
+}
+
 demand_setup<-function(setupDir, 
                        vista18TripsCsv,
                        out_weekday_activities_csv_gz,
                        out_weekend_activities_csv_gz,
                        out_weekday_activities_time_bins_csv_gz,
                        out_weekend_activities_time_bins_csv_gz,
-                       out_csv
+                       out_weekday_activities_end_time_dist_by_start_bins_csv_gz
                        ) {
   # example parameter values
   # setupDir <- '../output/1.setup'
@@ -14,7 +35,7 @@ demand_setup<-function(setupDir,
   # out_weekend_activities_csv_gz <- paste0(setupDir,'/vista_2012_18_extracted_activities_weekend.csv.gz')
   # out_weekday_activities_time_bins_csv_gz<-paste0(setupDir,'/vista_2012_18_extracted_activities_weekday_time_bins.csv.gz')
   # out_weekend_activities_time_bins_csv_gz<-paste0(setupDir,'/vista_2012_18_extracted_activities_weekend_time_bins.csv.gz')
-  # out_csv <- paste0(setupDir,'/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins.csv.gz')
+  # out_weekday_activities_end_time_dist_by_start_bins_csv_gz <- paste0(setupDir,'/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins.csv.gz')
 
   # Extract VISTA activities and save separately into weekday and weekend activities
   vista_csv <- vista18TripsCsv
@@ -42,8 +63,8 @@ demand_setup<-function(setupDir,
 
   # Write out the activity end time probabilities for each start time bin
   in_activities_csv_gz<-out_weekday_activities_csv_gz
-  echo(paste0('Extracting VISTA weekday activities end times distributions for each start time bin into ',out_csv,'\n'))
-  extract_and_write_activities_end_time_dist_by_start_bins(in_activities_csv_gz, out_csv, binsize)
+  echo(paste0('Extracting VISTA weekday activities end times distributions for each start time bin into ',out_weekday_activities_end_time_dist_by_start_bins_csv_gz,'\n'))
+  extract_and_write_activities_end_time_dist_by_start_bins(in_activities_csv_gz, out_weekday_activities_end_time_dist_by_start_bins_csv_gz, binsize)
 
   echo('Setup complete\n')
   return(TRUE)

diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_1.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_1.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_2.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_2.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_3.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_3.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_4.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_4.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_5.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_5.csv.gz
diff --git a/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_1.csv.gz b/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_1.csv.gz
diff --git a/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_2.csv.gz b/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_2.csv.gz
diff --git a/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_3.csv.gz b/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_3.csv.gz
diff --git a/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_4.csv.gz b/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_4.csv.gz
diff --git a/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_5.csv.gz b/...ected/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_5.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_1.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_1.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_2.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_2.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_3.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_3.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_4.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_4.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_5.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekday_time_bins_5.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_1.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_1.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_2.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_2.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_3.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_3.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_4.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_4.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_5.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_5.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_1.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_1.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_2.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_2.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_3.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_3.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_4.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_4.csv.gz
diff --git a/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_5.csv.gz b/tests/expected/1.setup/vista_2012_18_extracted_activities_weekend_time_bins_5.csv.gz
diff --git a/tests/testthat/testSetup.R b/tests/testthat/testSetup.R
@@ -7,27 +7,33 @@ test_that("VISTA 2012-18 trips pre-processing works", {
   outdir<-'../actual/1.setup'
   dir.create(outdir, showWarnings = FALSE, recursive=TRUE)
   capture_output(
-    demand_setup(
+    demand_setup_groups(
+      getGroupIds('../data/vistaCohorts.csv.gz'),
       outdir, 
       '../data/T_VISTA1218_V1.sample.csv',
-      '../actual/1.setup/vista_2012_18_extracted_activities_weekday.csv.gz',
-      '../actual/1.setup/vista_2012_18_extracted_activities_weekend.csv.gz',
-      '../actual/1.setup/vista_2012_18_extracted_activities_weekday_time_bins.csv.gz',
-      '../actual/1.setup/vista_2012_18_extracted_activities_weekend_time_bins.csv.gz',
-      '../actual/1.setup/vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins.csv.gz'
+      'vista_2012_18_extracted_activities_weekday_',
+      'vista_2012_18_extracted_activities_weekend_',
+      'vista_2012_18_extracted_activities_weekday_time_bins_',
+      'vista_2012_18_extracted_activities_weekend_time_bins_',
+      'vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_'
     )
   )
 
-  files<-c(
-    'vista_2012_18_extracted_activities_weekday.csv.gz',
-    'vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins.csv.gz',
-    'vista_2012_18_extracted_activities_weekday_time_bins.csv.gz',
-    'vista_2012_18_extracted_activities_weekend.csv.gz',
-    'vista_2012_18_extracted_activities_weekend_time_bins.csv.gz'
+  files_prefix<-c(
+    'vista_2012_18_extracted_activities_weekday_',
+    'vista_2012_18_extracted_activities_weekday_end_dist_for_start_bins_',
+    'vista_2012_18_extracted_activities_weekday_time_bins_',
+    'vista_2012_18_extracted_activities_weekend_',
+    'vista_2012_18_extracted_activities_weekend_time_bins_'
   )
-  for (file in files) {
-    expect_true(file.exists(paste0('../actual/1.setup/', file)))
-    expect_true(md5sum(paste0('../actual/1.setup/', file)) == md5sum(paste0('../expected/1.setup/', file)))
+
+  groups<-getGroupIds('../data/vistaCohorts.csv.gz')
+  for (gid in groups) {
+    for (prefix in files_prefix) {
+      file<-paste0(prefix,gid,".csv.gz")
+      expect_true(file.exists(paste0('../actual/1.setup/',file)))
+      expect_true(md5sum(paste0('../actual/1.setup/', file)) == md5sum(paste0('../expected/1.setup/', file)))
+    }
   }
 })