[fill-NA-values][xs] Filling missing values - refs #39

datasets · Nov 21, 2024 · 324c43e · 324c43e
1 parent 08b1422
commit 324c43e
Show file tree

Hide file tree

Showing 2 changed files with 48 additions and 47 deletions.
diff --git a/data/code-list.csv b/data/code-list.csv
@@ -23736,53 +23736,53 @@ x,GP,PAP,Pointe-à-Pitre,Pointe-a-Pitre,,XX,--3-----,2307,,1614N 06131W,
 ,MZ,VNX,Vilanculos,Vilanculos,,RQ,---4----,0901,,2200S 03516E,
 ,MZ,VJB,Xai-Xai,Xai-Xai,G,AI,---4----,0212,,2503S 03339E,
 ,MZ,ZBZ,Zambezia,Zambezia,P,RQ,--3--6--,1007,,1254S 03942E,
-,,AIW,Ai-Ais,Ai-Ais,,AI,---4----,0001,,,
-,,ADI,Arandis,Arandis,,AI,---4----,0001,,,
-,,VFG,Ariamsvlei,Ariamsvlei,KA,RL,--3-----,1601,,2807S 01949E,
-,,ZAR,Aussenkehr,Aussenkehr,,RL,-----6--,0901,,2821S 01723E,
-,,BQI,Bagani,Bagani,,AI,---4----,0001,,,
-,,GOG,Gobabis,Gobabis,,AI,---4----,0001,,,
-,,GFY,Grootfontein,Grootfontein,,AI,---4----,9601,,,
-,,HAL,Halali,Halali,,AI,---4----,0001,,,
-,,HTN,Henties Bay (Hentiesbaai),Henties Bay (Hentiesbaai),ER,RL,--3-----,1501,,2207S 01417E,
-,,KAS,Karasburg,Karasburg,,AI,---45---,9805,,,
-,,KRB,Karibib,Karibib,,RL,--3-----,0201,,2156S 01551E,
-,,KAT,Katima Mulilo,Katima Mulilo,CA,RL,--3-----,1707,,1730S 02416E,
-,,KMP,Keetmanshoop,Keetmanshoop,,AI,---4----,9601,,,
-,,LUD,Lüderitz,Luderitz,,AI,1--4----,0212,,2639S 01510E,
-,,MTH,Maltahohe,Maltahohe,HA,RQ,-----6--,1001,,2450S 01659E,
-,,MAR,Mariental,Mariental,,AI,----5---,9805,,,
-,,MQG,Midgard,Midgard,,AI,---4----,0001,,,
-,,OKU,Mokuti Lodge,Mokuti Lodge,,AI,---4----,0001,,,
-,,MJO,Mount Etjo Lodge,Mount Etjo Lodge,,AI,---4----,0001,,,
-,,MPA,Mpacha,Mpacha,,AI,---4----,0001,,,
-,,NNI,Namutoni,Namutoni,,AI,---4----,0001,,,
-,,WOR,Noordoewer,Noordoewer,KA,AI,--3-5---,0307,,2845S 01737E,
-,,OOZ,Okahandja,Okahandja,OD,RL,-----6--,1401,,2159S 01655E,
-,,OKF,Okaukuejo,Okaukuejo,,AI,---4----,0001,,,
-,,OMA,Omaruru,Omaruru,ER,RL,--3-----,0901,,2126S 01556E,
-,,OMG,Omega,Omega,,AI,---4----,0001,,,
-,,OND,Ondangwa,Ondangwa,,AI,---4----,0001,,,
-,,OGV,Ongava Game Reserve,Ongava Game Reserve,,AI,---4----,0001,,,
-,,ONG,Ongwediva,Ongwediva,ON,RL,--3-----,1901,,1746S 01545E,
-,,OPW,Opuwa,Opuwa,,AI,---4----,0001,,,
-,,OMD,Oranjemund,Oranjemund,,RQ,---4----,9012,,,
-,,OHI,Oshakati,Oshakati,,AI,---45---,9805,,,
-,,OSH,Oshikango,Oshikango,OW,RL,-----6--,1407,,1724S 01553E,
-,,OVO,Oshivelo,Oshivelo,OT,RL,--3-----,1707,,1837S 01710E,
-,,OTP,Otavi,Otavi,OD,RL,--3-----,1601,,1939S 01720E,
-,,OTJ,Otjiwarongo,Otjiwarongo,,AI,---45---,9805,,,
-,,OUT,Outjo,Outjo,KU,RL,--3-----,1901,,2007S 01609E,
-,,RUA,Ruacana,Ruacana,KU,RQ,1----6--,1001,,1726S 01421E,
-,,NDU,Rundu,Rundu,,AI,---4----,0001,,,
-,,SZM,Sesriem,Sesriem,,AI,---4----,0001,,,
-,,SWP,Swakopmund,Swakopmund,,AI,---45---,9805,,,
-,,TCY,Terrace Bay,Terrace Bay,,AI,---4----,0001,,,
-,,TSB,Tsumeb,Tsumeb,,AI,---4----,9601,,,
-,,UIS,Uis,Uis,ER,RL,--3-----,1801,,2113S 01451E,
-,,WVB,Walvis Bay,Walvis Bay,,QQ,1-------,8103,,,
-,,WDH,Windhoek,Windhoek,,AI,---45---,9601,,,
-,,ERS,Windhoek Eros Airport,Windhoek Eros Airport,KH,RL,---4----,1607,,2236S 01704E,
+,NA,AIW,Ai-Ais,Ai-Ais,,AI,---4----,0001,,,
+,NA,ADI,Arandis,Arandis,,AI,---4----,0001,,,
+,NA,VFG,Ariamsvlei,Ariamsvlei,KA,RL,--3-----,1601,,2807S 01949E,
+,NA,ZAR,Aussenkehr,Aussenkehr,,RL,-----6--,0901,,2821S 01723E,
+,NA,BQI,Bagani,Bagani,,AI,---4----,0001,,,
+,NA,GOG,Gobabis,Gobabis,,AI,---4----,0001,,,
+,NA,GFY,Grootfontein,Grootfontein,,AI,---4----,9601,,,
+,NA,HAL,Halali,Halali,,AI,---4----,0001,,,
+,NA,HTN,Henties Bay (Hentiesbaai),Henties Bay (Hentiesbaai),ER,RL,--3-----,1501,,2207S 01417E,
+,NA,KAS,Karasburg,Karasburg,,AI,---45---,9805,,,
+,NA,KRB,Karibib,Karibib,,RL,--3-----,0201,,2156S 01551E,
+,NA,KAT,Katima Mulilo,Katima Mulilo,CA,RL,--3-----,1707,,1730S 02416E,
+,NA,KMP,Keetmanshoop,Keetmanshoop,,AI,---4----,9601,,,
+,NA,LUD,Lüderitz,Luderitz,,AI,1--4----,0212,,2639S 01510E,
+,NA,MTH,Maltahohe,Maltahohe,HA,RQ,-----6--,1001,,2450S 01659E,
+,NA,MAR,Mariental,Mariental,,AI,----5---,9805,,,
+,NA,MQG,Midgard,Midgard,,AI,---4----,0001,,,
+,NA,OKU,Mokuti Lodge,Mokuti Lodge,,AI,---4----,0001,,,
+,NA,MJO,Mount Etjo Lodge,Mount Etjo Lodge,,AI,---4----,0001,,,
+,NA,MPA,Mpacha,Mpacha,,AI,---4----,0001,,,
+,NA,NNI,Namutoni,Namutoni,,AI,---4----,0001,,,
+,NA,WOR,Noordoewer,Noordoewer,KA,AI,--3-5---,0307,,2845S 01737E,
+,NA,OOZ,Okahandja,Okahandja,OD,RL,-----6--,1401,,2159S 01655E,
+,NA,OKF,Okaukuejo,Okaukuejo,,AI,---4----,0001,,,
+,NA,OMA,Omaruru,Omaruru,ER,RL,--3-----,0901,,2126S 01556E,
+,NA,OMG,Omega,Omega,,AI,---4----,0001,,,
+,NA,OND,Ondangwa,Ondangwa,,AI,---4----,0001,,,
+,NA,OGV,Ongava Game Reserve,Ongava Game Reserve,,AI,---4----,0001,,,
+,NA,ONG,Ongwediva,Ongwediva,ON,RL,--3-----,1901,,1746S 01545E,
+,NA,OPW,Opuwa,Opuwa,,AI,---4----,0001,,,
+,NA,OMD,Oranjemund,Oranjemund,,RQ,---4----,9012,,,
+,NA,OHI,Oshakati,Oshakati,,AI,---45---,9805,,,
+,NA,OSH,Oshikango,Oshikango,OW,RL,-----6--,1407,,1724S 01553E,
+,NA,OVO,Oshivelo,Oshivelo,OT,RL,--3-----,1707,,1837S 01710E,
+,NA,OTP,Otavi,Otavi,OD,RL,--3-----,1601,,1939S 01720E,
+,NA,OTJ,Otjiwarongo,Otjiwarongo,,AI,---45---,9805,,,
+,NA,OUT,Outjo,Outjo,KU,RL,--3-----,1901,,2007S 01609E,
+,NA,RUA,Ruacana,Ruacana,KU,RQ,1----6--,1001,,1726S 01421E,
+,NA,NDU,Rundu,Rundu,,AI,---4----,0001,,,
+,NA,SZM,Sesriem,Sesriem,,AI,---4----,0001,,,
+,NA,SWP,Swakopmund,Swakopmund,,AI,---45---,9805,,,
+,NA,TCY,Terrace Bay,Terrace Bay,,AI,---4----,0001,,,
+,NA,TSB,Tsumeb,Tsumeb,,AI,---4----,9601,,,
+,NA,UIS,Uis,Uis,ER,RL,--3-----,1801,,2113S 01451E,
+,NA,WVB,Walvis Bay,Walvis Bay,,QQ,1-------,8103,,,
+,NA,WDH,Windhoek,Windhoek,,AI,---45---,9601,,,
+,NA,ERS,Windhoek Eros Airport,Windhoek Eros Airport,KH,RL,---4----,1607,,2236S 01704E,
 ,NC,BDB,Baie de Prony,Baie de Prony,,RL,--3-----,0212,,2222S 16652E,
 ,NC,BMY,Bélep I.,Belep I.,,AI,---4----,1901,,1943S 16340E,
 ,NC,4NB,Bourail,Bourail,,RL,-----6--,1901,,2134S 16529E,

diff --git a/scripts/prepare.py b/scripts/prepare.py
@@ -113,6 +113,7 @@ def process(extracted_files):
     codelist_df = codelist_df.reindex(columns=['Change', 'Country', 'Location', 'Name', 'NameWoDiacritics', 'Subdivision',
                         'Status', 'Function', 'Date', 'IATA', 'Coordinates', 'Remarks'])
     # Keep only rows where 'Country' values are empty, 1 character, or exactly 2 characters
+    codelist_df['Country'] = codelist_df['Country'].fillna('NA')
     codelist_df = codelist_df[codelist_df['Country'].str.len().fillna(0).between(0, 2)]
     codelist_df = correct_swapped_function_status(codelist_df)
     codelist_df = clean_extra_rows(codelist_df)