3. Conflict data

• Accessing Conflict Data

We access conflict data from Uganda from the past 20 years. We clean and end up with a list of events, describing the actors involved, the type of conflict and its location.

Accessing Conflict Data

• Armed Conflict Location & Event Data Website
• API Guide
• Dictionary
• Codebook
• Full Documentation

Here, we are cleaning data from inidividual conflict events from the ACLED API.

%run /Users/thomasadler/Desktop/futuristic-platipus/capstone/notebooks/ta_01_packages_functions.py

/Users/thomasadler/opt/anaconda3/lib/python3.9/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.
  from pandas import MultiIndex, Int64Index

%run /Users/thomasadler/Desktop/futuristic-platipus/keys.py

conflict_api_endpoint = "https://api.acleddata.com/acled/read"

We want to get every single conflict event that happened in Uganda since 1997 (the start of the dataset).

uganda_iso = 800

conflict_r = requests.get(
    f'{conflict_api_endpoint}?key={conflict_api_key}&email={conflict_api_email}&limit=0&iso={uganda_iso}.csv'
)

#saving as json data
data = conflict_r.json()

#extract events information
events = data['data']

#save to a dataframe
uganda_conflict = pd.DataFrame(events)

uganda_conflict_df = uganda_conflict.copy()

uganda_conflict_df.tail()

	data_id	iso	event_id_cnty	event_id_no_cnty	event_date	year	time_precision	event_type	sub_event_type	actor1	...	location	latitude	longitude	geo_precision	source	source_scale	notes	fatalities	timestamp	iso3
7849	6876098	800	UGA5	5	1997-01-11	1997	1	Violence against civilians	Abduction/forced disappearance	LRA: Lords Resistance Army	...	Acholi-Bur	3.1258	32.9197	1	New York Times	International	LRA abduct an unknown number of people taking ...	0	1618581922	UGA
7850	6876117	800	UGA4	4	1997-01-08	1997	1	Battles	Armed clash	Military Forces of Uganda (1986-)	...	Kasese	0.1833	30.0833	3	Local Source	Subnational	Battle between Ugandan army and ADF rebels - 2...	2	1618581759	UGA
7851	6876122	800	UGA3	3	1997-01-07	1997	1	Battles	Armed clash	Military Forces of Uganda (1986-)	...	Nyabani	0.1358	30.3636	1	Local Source	Subnational	5 ADF rebels were killed when the Ugandan army...	5	1618581598	UGA
7852	6876154	800	UGA1	1	1997-01-01	1997	3	Battles	Armed clash	Military Forces of Uganda (1986-)	...	Gulu	2.7667	32.3056	3	Africa Research Bulletin	Other	Ugandan army battled with LRA rebels - 4 rebel...	4	1618581296	UGA
7853	6876155	800	UGA2	2	1997-01-01	1997	3	Battles	Armed clash	Military Forces of Uganda (1986-)	...	Mityana	0.4015	32.0452	3	Africa Research Bulletin	Other	Over 20 rebel groups believed to belong to Dun...	5	1618581439	UGA

5 rows × 31 columns

uganda_conflict_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7854 entries, 0 to 7853
Data columns (total 31 columns):
 #   Column            Non-Null Count  Dtype 
---  ------            --------------  ----- 
 0   data_id           7854 non-null   object
 1   iso               7854 non-null   object
 2   event_id_cnty     7854 non-null   object
 3   event_id_no_cnty  7854 non-null   object
 4   event_date        7854 non-null   object
 5   year              7854 non-null   object
 6   time_precision    7854 non-null   object
 7   event_type        7854 non-null   object
 8   sub_event_type    7854 non-null   object
 9   actor1            7854 non-null   object
 10  assoc_actor_1     7854 non-null   object
 11  inter1            7854 non-null   object
 12  actor2            7854 non-null   object
 13  assoc_actor_2     7854 non-null   object
 14  inter2            7854 non-null   object
 15  interaction       7854 non-null   object
 16  region            7854 non-null   object
 17  country           7854 non-null   object
 18  admin1            7854 non-null   object
 19  admin2            7854 non-null   object
 20  admin3            7854 non-null   object
 21  location          7854 non-null   object
 22  latitude          7854 non-null   object
 23  longitude         7854 non-null   object
 24  geo_precision     7854 non-null   object
 25  source            7854 non-null   object
 26  source_scale      7854 non-null   object
 27  notes             7854 non-null   object
 28  fatalities        7854 non-null   object
 29  timestamp         7854 non-null   object
 30  iso3              7854 non-null   object
dtypes: object(31)
memory usage: 1.9+ MB

num_columns = [
    'latitude',
    'longitude',
    'fatalities',
]

for col in num_columns:
    float_converter(uganda_conflict_df, col)

#check
uganda_conflict_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7854 entries, 0 to 7853
Data columns (total 31 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   data_id           7854 non-null   object 
 1   iso               7854 non-null   object 
 2   event_id_cnty     7854 non-null   object 
 3   event_id_no_cnty  7854 non-null   object 
 4   event_date        7854 non-null   object 
 5   year              7854 non-null   object 
 6   time_precision    7854 non-null   object 
 7   event_type        7854 non-null   object 
 8   sub_event_type    7854 non-null   object 
 9   actor1            7854 non-null   object 
 10  assoc_actor_1     7854 non-null   object 
 11  inter1            7854 non-null   object 
 12  actor2            7854 non-null   object 
 13  assoc_actor_2     7854 non-null   object 
 14  inter2            7854 non-null   object 
 15  interaction       7854 non-null   object 
 16  region            7854 non-null   object 
 17  country           7854 non-null   object 
 18  admin1            7854 non-null   object 
 19  admin2            7854 non-null   object 
 20  admin3            7854 non-null   object 
 21  location          7854 non-null   object 
 22  latitude          7854 non-null   float32
 23  longitude         7854 non-null   float32
 24  geo_precision     7854 non-null   object 
 25  source            7854 non-null   object 
 26  source_scale      7854 non-null   object 
 27  notes             7854 non-null   object 
 28  fatalities        7854 non-null   float32
 29  timestamp         7854 non-null   object 
 30  iso3              7854 non-null   object 
dtypes: float32(3), object(28)
memory usage: 1.8+ MB

date_converter(uganda_conflict_df, 'event_date')

#check
uganda_conflict_df['event_date']

0      2022-07-28
1      2022-07-28
2      2022-07-27
3      2022-07-27
4      2022-07-27
          ...    
7849   1997-01-11
7850   1997-01-08
7851   1997-01-07
7852   1997-01-01
7853   1997-01-01
Name: event_date, Length: 7854, dtype: datetime64[ns]

uganda_conflict_df=pd.DataFrame(uganda_conflict_df.drop(columns=['time_precision', 'event_id_cnty','event_id_no_cnty',\
                                                              'geo_precision','timestamp','year',\
                                                             'iso','iso3', 'region','country']))

#check current columns
uganda_conflict_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7854 entries, 0 to 7853
Data columns (total 21 columns):
 #   Column          Non-Null Count  Dtype         
---  ------          --------------  -----         
 0   data_id         7854 non-null   object        
 1   event_date      7854 non-null   datetime64[ns]
 2   event_type      7854 non-null   object        
 3   sub_event_type  7854 non-null   object        
 4   actor1          7854 non-null   object        
 5   assoc_actor_1   7854 non-null   object        
 6   inter1          7854 non-null   object        
 7   actor2          7854 non-null   object        
 8   assoc_actor_2   7854 non-null   object        
 9   inter2          7854 non-null   object        
 10  interaction     7854 non-null   object        
 11  admin1          7854 non-null   object        
 12  admin2          7854 non-null   object        
 13  admin3          7854 non-null   object        
 14  location        7854 non-null   object        
 15  latitude        7854 non-null   float32       
 16  longitude       7854 non-null   float32       
 17  source          7854 non-null   object        
 18  source_scale    7854 non-null   object        
 19  notes           7854 non-null   object        
 20  fatalities      7854 non-null   float32       
dtypes: datetime64[ns](1), float32(3), object(17)
memory usage: 1.2+ MB

print(
    'admin1 in the conflict dataset should be clean_adm1 in the water dataset, check with:',
    uganda_conflict_df['admin1'].head(1)[0])
print(
    'admin2 in the conflict dataset should be clean_adm2 in the water dataset, check with:',
    uganda_conflict_df['admin2'].head(1)[0])
print(
    'admin3 in the conflict dataset should be clean_adm3 in the water dataset, check with:',
    uganda_conflict_df['admin3'].head(1)[0])
print(
    'location in the conflict dataset should be clean_adm4 in the water dataset, check with:',
    uganda_conflict_df['location'].head(1)[0])

admin1 in the conflict dataset should be clean_adm1 in the water dataset, check with: Northern
admin2 in the conflict dataset should be clean_adm2 in the water dataset, check with: Napak
admin3 in the conflict dataset should be clean_adm3 in the water dataset, check with: Bokora
location in the conflict dataset should be clean_adm4 in the water dataset, check with: Kalokengel

uganda_conflict_df.rename(columns={
    'admin1': 'clean_adm1',
    'admin2': 'clean_adm2',
    'admin3': 'clean_adm3',
    'location': 'clean_adm4'
},
                          inplace=True)

print(uganda_conflict_df.isna().sum().sum()>0,\
      uganda_conflict_df.duplicated().sum()>0,\
      uganda_conflict_df.T.duplicated().sum()>0)

False False False

uganda_conflict_df.to_csv(data_filepath + 'ta_3_conflict_df_clean.csv')

Image(dictionary_filepath+"3A-Conflict-Dictionary.png")

uganda_conflict_df_upper = uganda_conflict_df.copy()

for col in['clean_adm1', 'clean_adm2', 'clean_adm3', 'clean_adm4']:
    uganda_conflict_df_upper[col] = uganda_conflict_df_upper[col].str.upper()

#export to cleaned dataset to csv
uganda_conflict_df_upper.to_csv(data_filepath + 'ta_3_conflict_df_clean_upper.csv')