COVID-19 analysis code 02 - data loading

covid19_analysis_02_data_loading.py

from datetime import datetime

csv_files = glob.glob(DATA_DIR + "/*.csv")

# Auxiliary function to extract date from file name
def extract_date(file_name):
    date_str = os.path.basename(file_name)[:-4]
    
    date = datetime.strptime(date_str, '%m-%d-%Y').date()
    
    return date

# Create dataframe
world_df = None

fields_mapping_dic = {
    "Province/State":"Province_State",
    "Country/Region":"Country_Region",
    "Last Update":"Last_Update",
    "Confirmed":"Confirmed",
    "Deaths":"Deaths",
    "Recovered":"Recovered",
    "Latitude":"Lat",
    "Longitude":"Long_"
}

for csv_file in csv_files:
    
    df = pd.read_csv(csv_file)
    date = extract_date(csv_file)
    
    # Data after 23/03/2020 must be adjusted due to change in data format
    if date >= datetime.strptime("03-23-2020", '%m-%d-%Y').date():
        for k, v in fields_mapping_dic.items():
            df[k] = df[v]
    
    df['Date'] = date
    
    if world_df is None:
        world_df = df
    else:
        world_df = pd.concat((world_df, df), ignore_index=True)
        
world_df.tail()