deidentified data

VCG · Oct 21, 2023 · 5e6091f · 5e6091f
1 parent 659336e
commit 5e6091f
Show file tree

Hide file tree

Showing 16 changed files with 3,103 additions and 3,074 deletions.
diff --git a/study_data/deidentify.py b/study_data/deidentify.py
@@ -0,0 +1,29 @@
+import pandas as pd
+import os
+
+
+paths_to_clean = [ 
+    os.path.join('full_Study','complete_study_mar16.csv'), 
+    os.path.join('full_Study','study.csv'),
+    os.path.join('full_Study','data_patch.csv'),
+    os.path.join('full_Study','data_clean.csv'),
+    os.path.join('full_Study','data_clean_june30.csv'),
+    os.path.join('full_Study','condition_count.csv'),
+    os.path.join('pilot2','Complexity vs. Trust in Vis_March 6, 2023_07.54.csv'),
+    os.path.join('pilot3','pilot3.csv'),
+    os.path.join('pilot3','pilot3_clean.csv'),
+    os.path.join('pilot3','Complexity vs. Trust in Vis_March 12, 2023_16.46.csv'),
+    os.path.join('pilot3','Complexity vs. Trust in Vis_March 12, 2023_16.47.csv'),
+    os.path.join('pilot4','data.csv'),
+    os.path.join('pilot4','data_patch.csv'),
+    os.path.join('replicationStudy','data.csv'),
+    os.path.join('replicationStudy','data_clean.csv')
+]
+for path in paths_to_clean:
+    df = pd.read_csv(path)
+    print('Interview_1_TEXT' in df.columns)
+    df = df.drop('Interview_1_TEXT',axis=1)
+    df.to_csv(path,index=False)
+# for col in df.columns:
+#     print(col)
+# print(df['Interview_1_TEXT'])
diff --git a/study_data/full_Study/complete_study_mar16.csv b/study_data/full_Study/complete_study_mar16.csv
diff --git a/study_data/full_Study/condition_count.csv b/study_data/full_Study/condition_count.csv
diff --git a/study_data/full_Study/data_clean.csv b/study_data/full_Study/data_clean.csv
diff --git a/study_data/full_Study/data_clean_june30.csv b/study_data/full_Study/data_clean_june30.csv
diff --git a/study_data/full_Study/data_patch.csv b/study_data/full_Study/data_patch.csv
diff --git a/study_data/full_Study/study.csv b/study_data/full_Study/study.csv
diff --git a/study_data/pilot2/Complexity vs. Trust in Vis_March 6, 2023_07.54.csv b/study_data/pilot2/Complexity vs. Trust in Vis_March 6, 2023_07.54.csv
diff --git a/study_data/pilot3/Complexity vs. Trust in Vis_March 12, 2023_16.46.csv b/study_data/pilot3/Complexity vs. Trust in Vis_March 12, 2023_16.46.csv
diff --git a/study_data/pilot3/Complexity vs. Trust in Vis_March 12, 2023_16.47.csv b/study_data/pilot3/Complexity vs. Trust in Vis_March 12, 2023_16.47.csv
diff --git a/study_data/pilot3/pilot3.csv b/study_data/pilot3/pilot3.csv
diff --git a/study_data/pilot3/pilot3_clean.csv b/study_data/pilot3/pilot3_clean.csv
diff --git a/study_data/pilot4/data.csv b/study_data/pilot4/data.csv
diff --git a/study_data/pilot4/data_patch.csv b/study_data/pilot4/data_patch.csv
diff --git a/study_data/replicationStudy/data.csv b/study_data/replicationStudy/data.csv
diff --git a/study_data/replicationStudy/data_clean.csv b/study_data/replicationStudy/data_clean.csv