From 0549d0f55fc4fdf453923a347219d7cc08db492c Mon Sep 17 00:00:00 2001
From: Utku Bilen Demir <84389167+UtkuBilenDemir@users.noreply.github.com>
Date: Mon, 24 Apr 2023 13:51:06 +0200
Subject: [PATCH] Check institution frequencies

---
 .../u_institution_frequency_exploration.py        |  14 ++++++++++++++
 .../wos_pubunique-institution_frequency.csv       |   3 +++
 .../~$wos_institution_locations_harmonized.xlsx   | Bin 0 -> 165 bytes
 3 files changed, 17 insertions(+)
 create mode 100644 WOS/wos_analysis/u_institution_frequency_exploration.py
 create mode 100644 WOS/wos_processed_data/wos_pubunique-institution_frequency.csv
 create mode 100644 WOS/wos_processed_data/~$wos_institution_locations_harmonized.xlsx

diff --git a/WOS/wos_analysis/u_institution_frequency_exploration.py b/WOS/wos_analysis/u_institution_frequency_exploration.py
new file mode 100644
index 0000000..5613e5d
--- /dev/null
+++ b/WOS/wos_analysis/u_institution_frequency_exploration.py
@@ -0,0 +1,14 @@
+#%% Compare the frequency of the inst. names with the wos-analysis on the web interface
+import pandas as pd
+
+# %% 
+wos_inst_df = pd.read_excel("../wos_processed_data/wos_institution_locations_harmonized.xlsx")
+
+# %% Get rid of the duplicate institution name entries in each individual publication
+wos_inst_df["id_inst"] = wos_inst_df['Institution'] + ";" + wos_inst_df["UT (Unique WOS ID)"]
+wos_pubunique_inst = wos_inst_df["id_inst"].unique()
+wos_pubunique_inst = [x.split(";")[0] for x in wos_pubunique_inst]
+
+# %% Calc. frequencies, and get an output
+wos_inst_freqdf = pd.DataFrame(wos_pubunique_inst).value_counts().rename_axis('institution').reset_index(name='frequency')
+wos_inst_freqdf.to_csv("../wos_processed_data/wos_pubunique-institution_frequency.csv", index=False)
diff --git a/WOS/wos_processed_data/wos_pubunique-institution_frequency.csv b/WOS/wos_processed_data/wos_pubunique-institution_frequency.csv
new file mode 100644
index 0000000..a175556
--- /dev/null
+++ b/WOS/wos_processed_data/wos_pubunique-institution_frequency.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d4a58f241d2e9b130869eedae2119be43919952d6dd236cdcd4d613a48d3e6b
+size 453006
diff --git a/WOS/wos_processed_data/~$wos_institution_locations_harmonized.xlsx b/WOS/wos_processed_data/~$wos_institution_locations_harmonized.xlsx
new file mode 100644
index 0000000000000000000000000000000000000000..a20b14c8f8c665d4246ab607a6b78dae4140ec16
GIT binary patch
literal 165
zcmd-LEy*rbaMDw7NzKhHQXm%aGK4aeFk~~7GAJ-OG3WtF7lu@ZT!u`BA|RXE5CFxp
B67K*2

literal 0
HcmV?d00001