test_cities.py

import unittest
import pandas as pd

# Function to create the DataFrame
def create_cities_dataframe():
    cities_data = {
        'City': ["Tokyo", "Delhi", "Shanghai", "Dhaka", "São Paulo", "Mexico City", "Cairo", "Beijing", "Mumbai", "Osaka"],
        'Country': ["Japan", "India", "China", "Bangladesh", "Brazil", "Mexico", "Egypt", "China", "India", "Japan"],
        'Population (Millions)': [37.2, 32.9, 24.8, 23.2, 22.6, 22.3, 22.2, 21.9, 20.7, 19.1]
    }
    return pd.DataFrame(cities_data)

class TestCitiesDataFrame(unittest.TestCase):
    def setUp(self):
        self.df = create_cities_dataframe()

    def test_column_names(self):
        expected_columns = ['City', 'Country', 'Population (Millions)']
        self.assertListEqual(list(self.df.columns), expected_columns)

    def test_number_of_rows(self):
        self.assertEqual(len(self.df), 10)

    def test_population_values(self):
        expected_populations = [37.2, 32.9, 24.8, 23.2, 22.6, 22.3, 22.2, 21.9, 20.7, 19.1]
        self.assertListEqual(list(self.df['Population (Millions)']), expected_populations)

    def test_city_names(self):
        expected_cities = ["Tokyo", "Delhi", "Shanghai", "Dhaka", "São Paulo", "Mexico City", "Cairo", "Beijing", "Mumbai", "Osaka"]
        self.assertListEqual(list(self.df['City']), expected_cities)

if __name__ == '__main__':
    unittest.main()