将不同元素类型的列表转换为整数列表

Question

I have a list of different element types (extracted from a column from a dataframe) that I would like to convert to the same element type (integers). The dataframe looks like this:

Because some rows under column "Systemic Banking Crisis (starting date)" only have one year, while others have several, the extracted list ends up looking like this:

[1994, 1990, nan, '1980, 1989, 1995, 2001', 1994, nan, 2008, 1995, 1987, nan, 1995, 2008, nan,...]

The countries that have multiple years (multiple banking crises) are in a string, while the countries with only one year are a integer. I would like to turn the data into panel data by looping through each country and making a dummy variable running from 1970 to 2019 that takes the value 1 if there is a banking crisis and 0 if not. To do this I have run the following code:

data_banking = data['Systemic Banking Crisis (starting date)'].to_list()
data_currency = data['Currency Crisis (year)'].to_list()
countries = data['Country'].to_list()

#making lists
years = [1970]
for i in range(1971, 2020):
    years.append(i)
banking_crisis = []
currency_crisis = []

countries_long = []
for i in countries:
    country = [i for x in range(50)]
    countries_long.extend(country)

years_long = []
for i in range(166):
    years_long.extend(years)

    
for i in data_banking:
    for y in years:
        if y==i:
            banking_crisis.append(1)
        else:
            banking_crisis.append(0)
            
banking = pd.DataFrame(list(zip(countries_long, years_long, banking_crisis)))

This works for all the countries with only one banking crisis and returns a dataframe that looks like this:

However, for the countries with multiple banking crises, python doesn't understand the code because the years are in one string. How do I fix this? I have tried to convert the list data_banking to a list of lists, convert all list elements to strings, then split the strings and convert each string element to integers, so that I could loop through each element in each (country)list of the data_banking list, but it won't work. These are the different variations of what I have tried:

def list_of_lists(lst):
    list_1 = [[el] for el in lst]
        #listToStr = ' '.join(map(str, lists))
    return list_1
        #list_1 = listToString(lists)
        #for string in list_values:
         #   list_values = list_1.split(",")
          #  string = int(string)
    #return list_1

data_banking = list_of_lists(data_banking)
for lists in data_banking:
    for item in lists:
        item = float(item)
#    lists = [str(x) for x in lists]

What should I do?

Answer 1

I'd do this entire operation in two steps. (1) First, I iterate over the dataset and store a list of dictionaries containing the country and each singular year its associated with (dropping NaNs), via some string formatting. (2) I then compile these results into a new data frame, making sure that the year column is numeric. Here's the code:

# Step 1
bank = 'Systemic Banking Crisis (starting date)'
rows = []
for _, row in data.iterrows():
    country = row['Country']
    years = row[bank]
    if pd.isna(years):
        continue
    for year in years.split(','):
        rows.append({'Country': country, bank:pd.to_numeric(year)})
# Step 2
df = pd.DataFrame(rows)
df[bank] pd.to_numeric(df[bank])

Let me know if this doesn't work for you.