def test_unequal_length_in_list():
"""
Test if assertion error will be thrown if the lists have different length
"""
with pytest.raises(AssertionError, match=r'.* empty list .*'):
get_distance(point1, point3)
get_distance(point3, point1)
def test_null_list_input():
"""
Test if the Value error will be raised if one of the parameter is empty list
"""
with pytest.raises(ValueError, match=r'.* empty list .*'):
get_distance(point1, empty_point)
get_distance(empty_point, point1)
def test_correct_cosine():
"""
Test if the correct distance is return based on the metric
"""
assert get_distance(point1, point2,
metric="eclidean") == distance.cosine(point1, point2)
def get_all_distances(point, data, dist="euclidean"):
"""
Return distance metric for each row in a dataframe as compared to an input list
Compares an input reference vector to all rows of an input data frame, calculating the specified distance/similarity metric for each row.
Parameters
----------
data : pandas dataframe
dataframe of size n by k to compare to point
point: list
list of length k to compare to data
dist: string
string indicating type of distance metric
Returns
-------
list
numeric vector of length n containing distances for each row of data
Example
-------
df = pd.DataFrame({"A" : [1,2,3], "B" : [8,2,4]})
point = [-2,4]
get_all_distances(point, df, dist = "euclidean")
>>> [5, 4.47, 5]
"""
# raise error if dataframe isn't the correct type of object
if not isinstance(data, pd.DataFrame):
raise Exception("the data argument should be a pandas dataframe")
# raise error if first argument isn't a list
if not isinstance(point, list):
raise Exception("the point argument should be type list")
# number of observations in data frame
n = data.shape[0]
k = data.shape[1]
# raise error if point isn't length k
if len(point) != k:
raise Exception(
"point vector length and number of columns in data should match")
# raise error if dist isn't correctly defined
if not dist in ["euclidean", "cosine", "manhattan"]:
raise Exception(
"dist should be one of 'euclidean','cosine' or 'manhattan'")
# empty vector to be filled with distances
distances = []
for obs in range(0, n):
distances.append(get_distance(point, data.iloc[obs, ], metric=dist))
return distances
def test_correct_manhattan():
"""
Test if the correct distance is return based on the metric
"""
assert get_distance(point1, point2,
metric="manhattan") == get_manhattan_dist(
point1, point2)
def test_non_numeric_element_output():
"""
Test if the Value error will be raised if one of the parameter is empty list
"""
with pytest.raises(ValueError, match=r'.*non-numeric value.*'):
get_distance(point1, point2, testing='output')
def test_non_numeric_element_input():
"""
Test if the Value error will be raised if one of the parameter has non-numeric
"""
with pytest.raises(ValueError, match=r'.*non-numeric element.*'):
get_distance(point1, bad_point)
def test_incorrect_metric():
"""
Test if the metric provided is correct
"""
with pytest.raises(KeyError, match=r'.*metric has to be one of'):
get_distance(point1, point2, metric="error test")