data/population_based/_20pm4g_.py

def true_code():
    import numpy as np
    import pandas as pd
    main_data = pd.read_csv("raw_data/main_data.csv")
    main_data['Timestamp'] = pd.to_datetime(main_data['Timestamp'])
    states_data = pd.read_csv("raw_data/State_data.csv")
    ncap_funding_data = pd.read_csv("raw_data/NCAP_Funding.csv")
    ncap_funding_data.replace('-', np.nan, inplace=True)
    ncap_funding_data['Amount released during FY 2019-20'] = ncap_funding_data['Amount released during FY 2019-20'].astype('float64')
    ncap_funding_data['Amount released during FY 2020-21'] = ncap_funding_data['Amount released during FY 2020-21'].astype('float64')
    ncap_funding_data['Amount released during FY 2021-22'] = ncap_funding_data['Amount released during FY 2021-22'].astype('float64')
    ncap_funding_data['Utilisation as on June 2022'] = ncap_funding_data['Utilisation as on June 2022'].astype('float64')
    
    state_funding = ncap_funding_data.groupby('state')['Total fund released'].sum().reset_index()
    merged_data = pd.merge(states_data, state_funding, on='state')
    merged_data['funding_per_capita'] = merged_data['Total fund released'] / merged_data['population']
    median_population = states_data['population'].median()
    low_pop_states = merged_data[merged_data['population'] < median_population]
    max_funding_state = low_pop_states.loc[low_pop_states['funding_per_capita'].idxmax()]
    print(max_funding_state['state'])

true_code()