[FIXED] Calculate the growth ratio from first value in every column in a pandas dataframe and return a Numpy array

Issue

need a bit of help with my function.

This is what I'm trying to do:

Build a predictive model that can give us the best guess at what the population growth rate in a given year might be. We will calculate the population growth rate as follows:

As such, we can only calculate the growth rate for the year 1961 onwards.

Write a function that takes the population_df and a country_code as input and computes the population growth rate for a given country starting from the year 1961. This function must return a return a 2-d numpy array that contains the year and corresponding growth rate for the country.

Function Specifications:

Should take a population_df and country_code string as input and return a numpy array as output. The array should only have two columns containing the year and the population growth rate, in other words, it should have a shape (?, 2) where ? is the length of the data.

𝐺𝑟𝑜𝑤𝑡ℎ_𝑟𝑎𝑡𝑒 = 𝑐𝑢𝑟𝑟𝑒𝑛𝑡_𝑦𝑒𝑎𝑟_𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 − 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠_𝑦𝑒𝑎𝑟_𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 / 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠_𝑦𝑒𝑎𝑟_𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛

Should take a population_df and country_code string as input and return a numpy array as output. The array should only have two columns containing the year and the population growth rate, in other words, it should have a shape (?, 2) where ? is the length of the data.

Input DF head:

My code:(Changeable)

def pop_growth_by_country_year(df,country_code):
    country_data = df.loc[country_code]
    for columnName, columnData in country_data.iteritems():
        country_data = ((country_data[columnData] - country_data[columnData-1]) / country_data[columnData-1])
    output = country_data.reset_index().to_numpy().reshape(-1, 2)
    return output

Input function(Not changeable)

pop_growth_by_country_year(population_df,'ABW')

Expected output:

array([[ 1.961e+03,  2.263e-02],
       [ 1.962e+03,  1.420e-02],
       [ 1.963e+03,  8.360e-03],
       [ 1.964e+03,  5.940e-03],
            ...       ....
       [ 2.015e+03,  5.260e-03],
       [ 2.016e+03,  4.610e-03],
       [ 2.017e+03,  4.220e-03]])

Solution

My input:

population_df = pd.DataFrame({
    '1960': {'ABW': 54211.0, 'AFG': 8996351.0, 'AGO': 5643182.0, 'ALB': 1608800.0, 'AND': 13411.0},
    '1961': {'ABW': 55438.0, 'AFG': 9166764.0, 'AGO': 5753024.0, 'ALB': 1659800.0, 'AND': 14375.0},
    '1962': {'ABW': 56225.0, 'AFG': 9345868.0, 'AGO': 5866061.0, 'ALB': 1711319.0, 'AND': 15370.0},
    '1963': {'ABW': 56695.0, 'AFG': 9533954.0, 'AGO': 5980417.0, 'ALB': 1762621.0, 'AND': 16412.0}
})
population_df

My solution:

def pop_growth_by_country_year(df,country_code):
    current_population = df.loc[country_code]
    previous_population = current_population.shift(1)
    growth = (current_population-previous_population)/previous_population
    return growth.dropna().reset_index().astype(float).values

Output of pop_growth_by_country_year(population_df,'ABW')

array([[1.96100000e+03, 2.26337828e-02],
       [1.96200000e+03, 1.41960388e-02],
       [1.96300000e+03, 8.35927079e-03]])

Note that, since you don't have the previous population for the first year (1960 in this case), you will miss the growth for that year and for this reason len(output)=len(input)-1

Answered By - Salvatore Daniele Bianco