Why does django Unittest throw an assertion error when comparing two instances of numpy.float64?

Question

I'm writing some unit tests, one of which checks that the data provided in a dataframe is of the correct type (float).

When I run the test assertIsInstance(type(foo), np.float64) the test fails with the following error message: AssertionError <class numpy.float64> is not an instance of <class numpy.float64>

I would have expected this to pass.

test_dataframe_functions.py

import numpy as np
from django.test import TestCase
import pandas as pd
from myapp import dataframe_functions

class DataframeFunctionsTest(TestCase):
    dataframe_to_test = dataframe_functions.create_dataframe

    #passes
    def test_dataframe_is_definitely_a_dataframe(self):
        self.assertIsInstance(self.dataframe_to_test, pd.DataFrame)

    #passes
    def test_dataframe_has_the_right_columns(self):
        column_headers = list(self.dataframe_to_test.columns)
        self.assertEquals(column_headers, ['header1', 'header2', 'header3'])

    #fails with AssertionError <class numpy.float64> is not an instance of <class numpy.float64>
    def test_dataframe_header1_is_correct_format(self):
        data_instance = self.dataframe_to_test['header1'].iloc[1]
        self.assertIsInstance(type(data_instance), np.float64)

I've checked that type(data_instance) does equal "class numpy.float64" with the following line of code:

print(type(dataframe_to_test['header1'].iloc[1]))

Answer 1

Because a type object is indeed not an instance of an np.float64. The assertIsInstance method should be called with assertIsInstance(object, type), and it thus checks if object is an instance of a (subtype of) type. A type object is not an instance of np.float64.

You thus should call this with:

assertIsInstance(foo, np.float64)

not with:

assertIsInstance(type(foo), np.float64)