test_imputation_exercises.py

import pandas as pd
import numpy as np
import pytest
from pandas.testing import assert_frame_equal
from pandas.testing import assert_series_equal

from scripts.imputation import impute, impute_min, impute_with_max
    
# EXERCISE 1

# add a happy test 
# how could your function be misused, edge cases

def test_impute_min_one_value():
    data = pd.Series([1.0, np.nan, 3.0])    # 1. Define some input data
    expected = pd.Series([1.0, 1.0, 3.0])   # 2. Define what is expected to happen
    actual = impute_min(data)  				  	       # 3. Run function and record what happens 
    assert_series_equal(expected, actual)   # 4. Make sure expected and actual are equal

def test_impute_max_one_value():
    data = pd.Series([1.0, np.nan, 3.0])    # 1. Define some input data
    expected = pd.Series([1.0, 3.0, 3.0])   # 2. Define what is expected to happen
    actual = impute_with_max(data)  				  	       # 3. Run function and record what happens
    assert_series_equal(expected, actual)   # 4. Make sure expected and actual are equal

@pytest.mark.skip(reason="it fails")
def test_impute_min_two_values():
    # exercise idea: use different arrays
    input_series = pd.Series([1, 2, 3, np.nan, 4, np.nan])
    expected_series = pd.Series([1.0, 2.0, 3.0, 1.0, 4.0, 1.0])
    output_series = impute_min(series=input_series)
    print(output_series)
    assert_series_equal(output_series, expected_series)
    assert False

def test_impute_all_nan():
    data = pd.Series([np.nan, np.nan, np.nan])    # 1. Define some input data
    expected = pd.Series([np.nan, np.nan, np.nan])   # 2. Define what is expected to happen
    actual = impute(data)  				  	       # 3. Run function and record what happens 
    assert_series_equal(expected, actual)   # 4. Make sure expected and actual are equal