validation/event_tracking/model_e_survival_calculations.py

# Import libraries
import pandas as pd

# Set file paths
input_file_path = '<YOUR_DATA_PATH>/EXAMPLE_STUDY_DATA/'
output_file_path = '<YOUR_DATA_PATH>/summary_csv_files/'


def read_data(file):
    """
    Read in data source
    --------
    :param file: string filename
    :return: dataframe
    """
    df = pd.read_csv(file)
    return df


def format_data_for_output(survival_data):
    """
    Remove columns not needed for output
    --------
    :param survival_data: dataframe containing date of death field
    :return: above dataframe filtered to only contain columns required
    for analysis/ output
    """
    survival_data = survival_data[['SafeHavenID', 'DOD']]
    return survival_data


def filter_data(survival_data, date):
    """
    Filter data to only include those alive on the index date for analysis
    --------
    :param data: string filename
    :param date: Index date in 'DD-MM-YYYY' format
    :return: dataframe including only those alive on the index date for
    analysis
    """
    survival_data['DOD'] = pd.to_datetime(survival_data['DOD'])
    return survival_data[survival_data['DOD'] >= date]


def calulate_days_survived(survival_data, date):
    """
    Calcualte days survived following the index date
    --------
    :param survival_data: dataframe containing date of death field
    :param date: Index date in 'DD-MM-YYYY' format
    :return: days survived from index date
    """
    survival_data['index_date'] = date
    survival_data['index_date'] = pd.to_datetime(survival_data['index_date'])
    return (survival_data['DOD'] - survival_data['index_date']).dt.days


def main():
    # Read in data
    survival_file = input_file_path + "Deaths_Cohort3R.csv"
    survival_data = read_data(survival_file)

    # Drop duplicates
    survival_data = survival_data.drop_duplicates()

    # Filter to only include Safehaven ID and date of death fields
    survival_data = format_data_for_output(survival_data)

    # Filter data to only include deaths after the index date for analysis
    survival_data = filter_data(survival_data, '01-01-2020')

    # Calculate days survived
    survival_data['days_survived'] = calulate_days_survived(survival_data,
                                                            '01-01-2020')

    # Save data
    survival_data.to_pickle(output_file_path + 'Survival_from_index.pkl')


main()