def test_df_merging():
a = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
b = pam.DataFrame({"one": [10, 20, 30], "two": [20, 30, 40]})
assert a.append(b).values == [[1, 2], [2, 3], [3, 4], [10, 20], [20, 30], [30, 40]]
a = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
b = pam.DataFrame({"one": [10, 20, 30], "two": [20, 30, 40], "three": [0, 1, 1]})
assert a.append(b).values == [
[1, 2, nan],
[2, 3, nan],
[3, 4, nan],
[10, 20, 0],
[20, 30, 1],
[30, 40, 1],
]
assert b.append(a).values == [
[10, 20, 0],
[20, 30, 1],
[30, 40, 1],
[1, 2, nan],
[2, 3, nan],
[3, 4, nan],
]
assert a.append(b, ignore_index=False).index == (0, 1, 2, 0, 1, 2)
def test_groupby():
df = pam.DataFrame(
{
"Animal": ["Falcon", "Falcon", "Parrot", "Parrot"],
"Max Speed": [100, 120, 24, 26],
"Min Speed": [20, 22, 2, 4],
}
)
res = (
df.groupby("Animal").sum().loc[["Falcon", "Parrot"], ["Max Speed", "Min Speed"]]
)
assert res.values == [[220, 42], [50, 6]]
assert res.columns == ("Max Speed", "Min Speed")
assert res.index == ("Falcon", "Parrot")
res = (
df.groupby("Animal")
.mean()
.loc[["Falcon", "Parrot"], ["Max Speed", "Min Speed"]]
)
assert res.values == [[110.0, 21.0], [25.0, 3.0]]
assert res.columns == ("Max Speed", "Min Speed")
assert res.index == ("Falcon", "Parrot")
df = pam.DataFrame(
{
"Animal": ["Falcon", "Falcon", "Parrot", "Parrot"],
"Max Speed": [nan, nan, 24, 26],
"Min Speed": [20, 22, 2, 4],
}
)
res = (
df.groupby("Animal")
.apply(lambda x: sum(x))
.loc[["Falcon", "Parrot"], ["Max Speed", "Min Speed"]]
)
assert res.values == [[nan, 42], [50, 6]]
assert res.columns == ("Max Speed", "Min Speed")
assert res.index == ("Falcon", "Parrot")
res = (
df.groupby("Animal")
.apply(lambda x: sum(x), dropna=False)
.loc[["Falcon", "Parrot"], ["Max Speed", "Min Speed"]]
)
assert res.values == [[nan, 42], [50, 6]]
assert res.columns == ("Max Speed", "Min Speed")
assert res.index == ("Falcon", "Parrot")
def test_bound_ser():
a = pam.Series([0, 1, 2, 3])
assert a.bound_slice(slice(0, 3)) == slice(0, 3, 1)
assert a.bound_slice(slice(1, 100)) == slice(1, 4, 1)
assert a.bound_slice(slice(1, -1)) == slice(1, 3, 1)
assert a.bound_slice(slice(-3, -1)) == slice(1, 3, 1)
df = pam.DataFrame([[99, 98, 97, 97], [0, 1, 2, 3]])
a = df.iloc[1, :]
assert a.bound_slice(slice(0, 3)) == slice(1, 7, 2)
assert a.bound_slice(slice(1, 100)) == slice(3, 9, 2)
assert a.bound_slice(slice(1, -1)) == slice(3, 7, 2)
assert a.bound_slice(slice(-3, -1)) == slice(3, 7, 2)
a = df.iloc[:, 1]
assert a.bound_slice(slice(None, None)) == slice(2, 4, 1)
assert a.bound_slice(slice(None, 1)) == slice(2, 3, 1)
assert a.bound_slice(slice(None, 100)) == slice(2, 4, 1)
assert a.bound_slice(slice(None, -1)) == slice(2, 3, 1)
assert a.bound_slice(slice(None, -100)) == slice(2, 2, 1)
assert a.bound_slice(slice(1, None)) == slice(3, 4, 1)
assert a.bound_slice(slice(100, None)) == slice(4, 4, 1)
assert a.bound_slice(slice(-1, None)) == slice(3, 4, 1)
assert a.bound_slice(slice(-100, None)) == slice(2, 4, 1)
def test_is_2d_bool():
a = [True]
assert not pam.other_stuff.is_2d_bool(a)
a = [[True, False, True, True], [False, True, False, False]]
assert pam.other_stuff.is_2d_bool(a)
a = pam.DataFrame({"one": [True, True], "two": [True, True]})
assert pam.other_stuff.is_2d_bool(a)
def test_convert_slice():
df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 1, 2, 3, 4],
"two": [1, 2, 3, 4, 5],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
}
)
# try all on native dataframe
assert df.convert_slice(slice(None, None), axis=1) == slice(0, 5)
assert df.convert_slice(slice(1, None), axis=1) == slice(1, 5)
assert df.convert_slice(slice(1, -1), axis=1) == slice(1, 4)
assert df.convert_slice(slice(1, 2), axis=1) == slice(1, 2)
assert df.convert_slice(slice(None, None), axis=0) == slice(0, 5)
assert df.convert_slice(slice(1, None), axis=0) == slice(1, 5)
assert df.convert_slice(slice(1, -1), axis=0) == slice(1, 4)
assert df.convert_slice(slice(1, 2), axis=0) == slice(1, 2)
df = df.iloc[1:-1, 1:-1]
# df.view = (slice(1, 4), slice(1, 4))
assert df.convert_slice(slice(None, None), axis=1) == slice(0, 3)
assert df.convert_slice(slice(1, None), axis=1) == slice(1, 3)
assert df.convert_slice(slice(1, -1), axis=1) == slice(1, 2)
assert df.convert_slice(slice(1, 2), axis=1) == slice(1, 2)
assert df.convert_slice(slice(None, None), axis=0) == slice(0, 3)
assert df.convert_slice(slice(1, None), axis=0) == slice(1, 3)
assert df.convert_slice(slice(1, -1), axis=0) == slice(1, 2)
assert df.convert_slice(slice(1, 2), axis=0) == slice(1, 2)
def test_df_operators():
df = pam.DataFrame([[0, 10, 20], [1, 11, 21]])
assert (df > 10).values == [[False, False, True], [False, True, True]]
assert (~(df > 10)).values == [[True, True, False], [True, False, False]]
df1 = pam.DataFrame([[0, nan, 20], [1, 11, 21]])
df2 = pam.DataFrame([[0, nan, 20], [1, 11, 21]])
assert df1.equals(df2)
assert df2.equals(df1)
df2 = pam.DataFrame(
[[0, 0, 0, 0, 0], [0, 0, nan, 20, 0], [0, 1, 11, 21, 0], [0, 0, 0, 0, 0]]
)
assert not df1.equals(df2)
assert not df2.equals(df1)
df2 = df2.iloc[1:-1, 1:-1]
assert df1.equals(df2)
assert df2.equals(df1)
def test_bound_methods():
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 1, 2, 3, 4],
"two": [1, 2, 3, 4, 5],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
}
)
df = long_df.iloc[2:-1, 1:-1]
assert long_df.view == (slice(0, 5), slice(0, 5))
assert df.view == (slice(2, 4), slice(1, 4))
assert df.bound_slice_to_df(slice(None, None), axis=0) == slice(2, 4)
assert df.bound_slice_to_df(slice(None, 1), axis=0) == slice(2, 3)
assert df.bound_slice_to_df(slice(None, 100), axis=0) == slice(2, 4)
assert df.bound_slice_to_df(slice(None, -1), axis=0) == slice(2, 3)
assert df.bound_slice_to_df(slice(None, -100), axis=0) == slice(2, 2)
assert df.bound_slice_to_df(slice(1, None), axis=0) == slice(3, 4)
assert df.bound_slice_to_df(slice(1, 1), axis=0) == slice(3, 3)
assert df.bound_slice_to_df(slice(1, 100), axis=0) == slice(3, 4)
assert df.bound_slice_to_df(slice(1, -1), axis=0) == slice(3, 3)
assert df.bound_slice_to_df(slice(1, -100), axis=0) == slice(3, 2)
assert df.bound_slice_to_df(slice(100, None), axis=0) == slice(4, 4)
assert df.bound_slice_to_df(slice(100, 1), axis=0) == slice(4, 3)
assert df.bound_slice_to_df(slice(100, 100), axis=0) == slice(4, 4)
assert df.bound_slice_to_df(slice(100, -1), axis=0) == slice(4, 3)
assert df.bound_slice_to_df(slice(100, -100), axis=0) == slice(4, 2)
assert df.bound_int_to_df(0, axis=0) == 2
assert df.bound_int_to_df(-1, axis=0) == 3
assert df.bound_iterable_to_df([0, 1, -1], axis=0) == [2, 3, 3]
assert df.bound_slice_to_df(slice(None, None), axis=1) == slice(1, 4)
assert df.bound_slice_to_df(slice(None, 1), axis=1) == slice(1, 2)
assert df.bound_slice_to_df(slice(None, 100), axis=1) == slice(1, 4)
assert df.bound_slice_to_df(slice(None, -1), axis=1) == slice(1, 3)
assert df.bound_slice_to_df(slice(None, -100), axis=1) == slice(1, 1)
assert df.bound_slice_to_df(slice(1, None), axis=1) == slice(2, 4)
assert df.bound_slice_to_df(slice(1, 1), axis=1) == slice(2, 2)
assert df.bound_slice_to_df(slice(1, 100), axis=1) == slice(2, 4)
assert df.bound_slice_to_df(slice(1, -1), axis=1) == slice(2, 3)
assert df.bound_slice_to_df(slice(1, -100), axis=1) == slice(2, 1)
assert df.bound_slice_to_df(slice(100, None), axis=1) == slice(4, 4)
assert df.bound_slice_to_df(slice(100, 1), axis=1) == slice(4, 2)
assert df.bound_slice_to_df(slice(100, 100), axis=1) == slice(4, 4)
assert df.bound_slice_to_df(slice(100, -1), axis=1) == slice(4, 3)
assert df.bound_slice_to_df(slice(100, -100), axis=1) == slice(4, 1)
assert df.bound_int_to_df(0, axis=1) == 1
assert df.bound_int_to_df(-1, axis=1) == 3
assert df.bound_iterable_to_df([0, 1, -1], axis=1) == [1, 2, 3]
def test_init_dataframe():
# test dictionary, default index
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
assert df.index == (0, 1, 2)
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test dictionary, custom index
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]}, index=["a", "b", "c"])
assert df.index == ("a", "b", "c")
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test list of dicts, custom index
df = pam.DataFrame([{"one": [1, 2, 3]}, {"two": [2, 3, 4]}])
assert df.index == (0, 1, 2)
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test list of dicts, custom index
df = pam.DataFrame([{"one": [1, 2, 3]}, {"two": [2, 3, 4]}], index=["a", "b", "c"])
assert df.index == ("a", "b", "c")
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test rows
df = pam.DataFrame([[1, 2], [2, 3], [3, 4]])
assert df.index == (0, 1, 2)
assert df.columns == (0, 1)
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test rows with custom index
df = pam.DataFrame([[1, 2], [2, 3], [3, 4]], index=["a", "b", "c"])
assert df.index == ("a", "b", "c")
assert df.columns == (0, 1)
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test rows with custom index
df = pam.DataFrame([[1, 2], [2, 3], [3, 4]], columns=["one", "two"])
assert df.index == (0, 1, 2)
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test rows with custom index and column
df = pam.DataFrame(
[[1, 2], [2, 3], [3, 4]], columns=["one", "two"], index=["a", "b", "c"]
)
assert df.index == ("a", "b", "c")
assert df.columns == ("one", "two")
assert df.values == [[1, 2], [2, 3], [3, 4]]
def test_df_setitem_create():
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
},
index=[10, 11, 12, 13, 14],
)
df = long_df.iloc[1:-1, 1:-1]
df.loc[:, "dne"] = [99, 99, 99]
df.loc["dne", :] = 100
def test_df_getitem():
# tests for getting dataframes from non-view dataframes
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
},
index=[10, 11, 12, 13, 14],
)
exp_columns = ("one", "two", "three")
exp_index = (11, 12, 13)
exp_values = [[0, 1, 3], [1, 2, 4], [2, 3, 5]]
assert long_df["a"] == pam.Series([0, 0, 0, 0, 0], index=[10, 11, 12, 13, 14])
# non-view tests
df1 = long_df.iloc[1:-1, 1:-1]
assert df1.values == exp_values
assert df1.columns == exp_columns
assert df1.index == exp_index
df2 = long_df.iloc[1:-1, [1, 2, 3]]
assert df2.columns == exp_columns
assert df2.values == exp_values
assert df2.index == exp_index
df3 = long_df.iloc[[1, 2, 3], 1:-1]
assert df3.columns == exp_columns
assert df3.values == exp_values
assert df3.index == exp_index
df4 = long_df.iloc[[1, 2, 3], [1, 2, 3]]
assert df4.columns == exp_columns
assert df4.values == exp_values
assert df4.index == exp_index
# view tests
df = df1
exp_columns = ("two", "three")
exp_index = (12, 13)
exp_values = [[2, 4], [3, 5]]
df1 = df.iloc[1:, 1:]
assert df1.values == exp_values
assert df1.columns == exp_columns
assert df1.index == exp_index
df2 = df.iloc[1:, [1, 2]]
assert df2.columns == exp_columns
assert df2.values == exp_values
assert df2.index == exp_index
df3 = df.iloc[[1, 2], 1:]
assert df3.columns == exp_columns
assert df3.values == exp_values
assert df3.index == exp_index
df4 = df.iloc[[1, 2], [1, 2]]
assert df4.columns == exp_columns
assert df4.values == exp_values
assert df4.index == exp_index
# Tests for getting series from non-view dataframes
###
## Tests for dataframes without custom index
### test for view, int
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
ser = df.iloc[1:, 0]
assert ser.index == (1, 2)
assert ser.values == [2, 3]
assert all(ser == pam.Series([2, 3], index=[1, 2]))
## Tests for dtaframes with custom index
# test dataframe iloc matches a series with a custom index from a column
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]}, index=[10, 11, 12])
ser = df.iloc[1:, 0]
assert ser.index == (11, 12)
assert ser.values == [2, 3]
assert df.iloc[0, 1] == 2
assert all(ser == pam.Series([2, 3], index=[11, 12]))
# test dataframe iloc matches a series with a custom index from a column
df = pam.DataFrame(
{"one": [1, 2, 3], "two": [2, 3, 4], "three": [3, 4, 5], "four": [4, 5, 6]},
index=[10, 11, 12],
)
ser = df.iloc[0, 0:2]
assert ser.index == ("one", "two")
assert ser.values == [1, 2]
assert df.iloc[0, 1] == 2
assert all(ser == pam.Series([1, 2], index=["one", "two"]))
# test for view dataframes
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
}
)
df = long_df.iloc[1:-1, 1:-1]
ser = df.iloc[1:, 0]
assert ser.index == (2, 3)
assert ser.values == [1, 2]
assert ser.name == "one"
assert df.iloc[0, 1] == 1
assert all(ser == pam.Series([1, 2], index=[2, 3]))
# test for view dataframe
# non-custom index
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
}
)
df = long_df.iloc[1:-1, 1:-1]
ser = df.iloc[1:, 0]
assert ser.index == (2, 3)
assert ser.values == [1, 2]
assert ser.name == "one"
assert df.iloc[0, 1] == 1
assert all(ser == pam.Series([1, 2], index=[2, 3]))
long_df = pam.DataFrame(
{
"a": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [2, 3, 4, 5],
"b": [0, 0, 0, 0, 0],
},
index=[10, 11, 12, 13, 14],
)
df = long_df.iloc[1:-1, 1:-1]
ser = df.iloc[1:, 0]
assert ser.index == (12, 13)
assert ser.values == [1, 2]
assert ser.name == "one"
assert df.iloc[0, 1] == 1
assert all(ser == pam.Series([1, 2], index=[12, 13]))
# Test 2d Boolean arrays
# Test Boolean setitem
# DataFrame
exp_results = [[nan, nan, 100, 1000], [nan, nan, 200, 2000], [nan, 30, 300, 3000]]
input = [[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]]
df = pam.DataFrame(input)
df = df[df > 20]
assert df.values == exp_results
df = pam.DataFrame(input)
df = df.loc[df > 20]
assert df.values == exp_results
df = pam.DataFrame(input)
df = df.iloc[df > 20]
assert df.values == exp_results
# 2d boolean array
df = pam.DataFrame(input)
df = df[(df > 20).values]
assert df.values == exp_results
df = pam.DataFrame(input)
df = df.loc[(df > 20).values]
assert df.values == exp_results
df = pam.DataFrame(input)
df = df.iloc[(df > 20).values]
assert df.values == exp_results
def test_df_methods():
# test transpose
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
df2 = df.transpose()
assert df.equals(pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]}))
assert df.columns == ("one", "two")
assert df.index == (0, 1, 2)
assert df2.values == [[1, 2, 3], [2, 3, 4]]
assert df2.index == ("one", "two")
assert df2.columns == (0, 1, 2)
# test applymap
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
assert df.applymap(lambda x: x ** 2).values == [[1, 4], [4, 9], [9, 16]]
assert df.values == [[1, 2], [2, 3], [3, 4]]
# test apply
## non-reducing
### rows
df = pam.DataFrame({"one": [1, 2, 3], "two": [2, 3, 4]})
assert df.apply(lambda x: x ** 2, axis=1).values == [[1, 4], [4, 9], [9, 16]]
assert df.values == [[1, 2], [2, 3], [3, 4]]
### columns
assert df.apply(lambda x: x ** 2, axis=1).values == [[1, 4], [4, 9], [9, 16]]
assert df.values == [[1, 2], [2, 3], [3, 4]]
## reducing
def multiply(iterable):
res = reduce((lambda x, y: x * y), iterable)
return res
### rows
assert df.apply(multiply, axis=1).values == [2, 6, 12]
assert df.values == [[1, 2], [2, 3], [3, 4]]
### columns
assert df.apply(multiply, axis=0).values == [6, 24]
assert df.values == [[1, 2], [2, 3], [3, 4]]
### rows
assert df.apply(sum, axis=1).values == [3, 5, 7]
assert df.values == [[1, 2], [2, 3], [3, 4]]
### columns
assert df.apply(sum, axis=0).values == [6, 9]
assert df.values == [[1, 2], [2, 3], [3, 4]]
### with nas
df = pam.DataFrame({"one": [1, nan, 3], "two": [2, 3, 4]})
assert df.apply(sum, axis=0).values == [4, 9]
assert df.values == [[1, 2], [nan, 3], [3, 4]]
a = pam.DataFrame([[pam.nan, pam.nan, pam.nan], [1, 2, 3]])
assert a.apply(lambda x: sum(x), axis=1).values == [pam.nan, 6]
assert a.apply(lambda x: sum(x), axis=0).values == [1, 2, 3]
### columns
assert df.apply(multiply, axis=0).values == [3, 24]
assert df.values == [[1, 2], [nan, 3], [3, 4]]
# sort values
df = pam.DataFrame({"one": [1, 2, nan, 0, -1], "two": [nan, 10, nan, 0, -1]})
assert df.sort_values(by="one", axis=0).values == [
[-1, -1],
[0, 0],
[1, nan],
[2, 10],
[nan, nan],
]
assert df.sort_values(by="one", na_position="first", axis=0).values == [
[nan, nan],
[-1, -1],
[0, 0],
[1, nan],
[2, 10],
]
assert df.sort_values(by="one", ascending=False, axis=0).values == [
[2, 10],
[1, nan],
[0, 0],
[-1, -1],
[nan, nan],
]
assert df.sort_values(
by="one", ascending=False, na_position="first", axis=0
).values == [[nan, nan], [2, 10], [1, nan], [0, 0], [-1, -1]]
assert df.sort_values(
by="two", ascending=False, na_position="first", axis=0
).values == [[1, nan], [nan, nan], [2, 10], [0, 0], [-1, -1]]
assert df.equals(
pam.DataFrame({"one": [1, 2, nan, 0, -1], "two": [nan, 10, nan, 0, -1]})
)
df = pam.DataFrame({"one": [1, 2, nan, 0, 10], "two": [nan, 10, nan, 0, -1]})
assert df.sort_values(by=0, axis=1).values == [
[1, nan],
[2, 10],
[nan, nan],
[0, 0],
[10, -1],
]
assert df.sort_values(by=4, axis=1).values == [
[nan, 1],
[10, 2],
[nan, nan],
[0, 0],
[-1, 10],
]
assert df.equals(
pam.DataFrame({"one": [1, 2, nan, 0, 10], "two": [nan, 10, nan, 0, -1]})
)
# test reset_index
df = pam.DataFrame(
{"one": [1, 2, nan, 0, 10], "two": [nan, 10, nan, 0, -1]},
index=[10, 11, 12, 13, 14],
)
## test without drop
df2 = df.reset_index()
assert (
df.values
== pam.DataFrame(
{"one": [1, 2, nan, 0, 10], "two": [nan, 10, nan, 0, -1]},
index=[10, 11, 12, 13, 14],
).values
)
assert df2.values == [
[10, 1, nan],
[11, 2, 10],
[12, nan, nan],
[13, 0, 0],
[14, 10, -1],
]
assert df2.index == (0, 1, 2, 3, 4)
## test with drop
df2 = df.reset_index(drop=True)
assert (
df.values
== pam.DataFrame(
{"one": [1, 2, nan, 0, 10], "two": [nan, 10, nan, 0, -1]},
index=[10, 11, 12, 13, 14],
).values
)
assert df.equals(df2)
assert df2.index == (0, 1, 2, 3, 4)
## test iterrows()
data = [[1, nan], [2, 10], [nan, nan], [0, 0], [10, -1]]
row_names = [10, 11, 12, 13, 14]
columns = ["one", "two"]
df = pam.DataFrame(data, index=row_names, columns=columns)
rows = [(pam.Series(row, index=columns, name=j)) for row, j in zip(data, row_names)]
for row, row_name, row_tuple in zip(rows, row_names, df.iterrows()):
assert row_tuple[0] == row_name
assert row_tuple[1] == row
data = [[1, nan], [2, 10], [nan, nan], [0, 0], [10, -1]]
row_names = [10, 11, 12, 13, 14]
col_names = ["one", "two"]
df = pam.DataFrame(data, index=row_names, columns=col_names)
rows = [pam.Series(row, index=col_names, name=j) for row, j in zip(data, row_names)]
cols = [
pam.Series(col, index=row_names, name=j)
for col, j in zip(zip(*data), col_names)
]
for row, row_name, row_tuple in zip(rows, row_names, df.iterrows()):
assert row_tuple[0] == row_name
assert row_tuple[1] == row
for col, col_name, col_tuple in zip(cols, col_names, df.itercols()):
assert col_tuple[0] == col_name
assert col_tuple[1] == col
for col, col_name, col_tuple in zip(cols, col_names, df.iteritems()):
assert col_tuple[0] == col_name
assert col_tuple[1] == col
def test_series_methods():
# drop
a = pam.Series([0, 1, 2, 3])
a.drop(2)
# astype
a = pam.DataFrame([[1, 2, 4], [2, 3, 4]])
a.iloc[0] = a.iloc[0].astype(float)
assert a.values == [[1.0, 2.0, 4.0], [2, 3, 4]]
a = pam.DataFrame([[1, 2, 4], [2, 3, 4]])
a.iloc[:, 0] = a.iloc[:, 0].astype(float)
assert a.values == [[1.0, 2, 4], [2.0, 3, 4]]
a = pam.DataFrame([[1, 2, 4], [2, 3, 4]])
a.iloc[0] = a.iloc[0].apply(lambda x: x ** 2)
assert a.values == [[1, 4, 16], [2, 3, 4]]
a = pam.DataFrame([[1, 2, 4], [2, 3, 4]])
a.iloc[:, 0] = a.iloc[:, 0].apply(lambda x: x ** 2)
assert a.values == [[1, 2, 4], [4, 3, 4]]
# sort_values
ser = pam.Series([1, 10, 20, pam.other_stuff.nan, 40, 20, pam.other_stuff.nan])
assert ser.sort_values().values == [
1,
10,
20,
20,
40,
pam.other_stuff.nan,
pam.other_stuff.nan,
]
assert ser.sort_values(na_position="first").values == [
pam.other_stuff.nan,
pam.other_stuff.nan,
1,
10,
20,
20,
40,
]
assert ser.sort_values(ascending=False).values == [
40,
20,
20,
10,
1,
pam.other_stuff.nan,
pam.other_stuff.nan,
]
assert ser.sort_values(ascending=False, na_position="first").values == [
pam.other_stuff.nan,
pam.other_stuff.nan,
40,
20,
20,
10,
1,
]
# Test string
ser = pam.Series(["one", "two", "three"])
assert ser.str.upper().values == ["ONE", "TWO", "THREE"]
ser = pam.Series(
[
datetime.datetime(2007, 12, 6, 16, 29, 43),
datetime.datetime(2007, 12, 6, 16, 29, 43),
datetime.datetime(2007, 12, 6, 16, 29, 43),
]
)
# Test Datetime
assert ser.dt.replace(tzinfo=timezone.utc).values == [
datetime.datetime(2007, 12, 6, 16, 29, 43, tzinfo=datetime.timezone.utc),
datetime.datetime(2007, 12, 6, 16, 29, 43, tzinfo=datetime.timezone.utc),
datetime.datetime(2007, 12, 6, 16, 29, 43, tzinfo=datetime.timezone.utc),
]
assert ser == pam.Series(
[
datetime.datetime(2007, 12, 6, 16, 29, 43),
datetime.datetime(2007, 12, 6, 16, 29, 43),
datetime.datetime(2007, 12, 6, 16, 29, 43),
]
)
# Test isna
ser = pam.Series([1, 2, nan, 3])
assert ser.isna().values == [False, False, True, False]
# test dropna
print(pam.Series([1, 2, 3], index=(0, 1, 3)))
assert ser.dropna() == pam.Series([1, 2, 3], index=(0, 1, 3))
def test_ser_setitem():
# Test iloc
## test from series
ser = pam.Series([0, 1, 2, 3, 4], index=["zero", "one", "two", "three", "four"])
#### test integer
ser.iloc[0] = 9
assert ser.values == [9, 1, 2, 3, 4]
ser.iloc[4] = 99
assert ser.values == [9, 1, 2, 3, 99]
with pytest.raises(IndexError):
ser.iloc[6] = 100
#### test slice
ser.iloc[:] = [10, 11, 12, 13, 14]
assert ser.values == [10, 11, 12, 13, 14]
ser.iloc[:3] = 3
assert ser.values == [3, 3, 3, 13, 14]
ser.iloc[:5] = [99, 99, 99, 99, 99]
assert ser.values == [99, 99, 99, 99, 99]
#### test iterable
ser.iloc[[0, 1, 3]] = 100
assert ser.values == [100, 100, 99, 100, 99]
#### test bool
ser.iloc[[True, True, False, False, True]] = [0, 1, 4]
assert ser.values == [0, 1, 99, 100, 4]
ser.iloc[pam.Series([True, True, False, False, True])] = [10, 11, 14]
assert ser.values == [10, 11, 99, 100, 14]
# test from series view of a dataframe
## horizontal/row slice
df = pam.DataFrame(
[[11, 12, 13, 14, 15], [0, 1, 2, 3, 4]],
columns=["zero", "one", "two", "three", "four"],
)
ser = df.iloc[1, :]
#### test integer
ser.iloc[0] = 9
assert ser.values == [9, 1, 2, 3, 4]
ser.iloc[4] = 99
assert ser.values == [9, 1, 2, 3, 99]
with pytest.raises(IndexError):
ser.iloc[6] = 100
#### test slice
ser.iloc[:] = [10, 11, 12, 13, 14]
assert ser.values == [10, 11, 12, 13, 14]
ser.iloc[:3] = 3
assert ser.values == [3, 3, 3, 13, 14]
ser.iloc[:5] = [99, 99, 99, 99, 99]
assert ser.values == [99, 99, 99, 99, 99]
#### test iterable
ser.iloc[[0, 1, 3]] = 100
assert ser.values == [100, 100, 99, 100, 99]
#### test bool
ser.iloc[[True, True, False, False, True]] = [0, 1, 4]
assert ser.values == [0, 1, 99, 100, 4]
ser.iloc[pam.Series([True, True, False, False, True])] = [10, 11, 14]
assert ser.values == [10, 11, 99, 100, 14]
## vertical/column slice
df = pam.DataFrame(
[[11, 12, 13, 14, 15], [0, 1, 2, 3, 4]],
columns=["zero", "one", "two", "three", "four"],
)
ser = df.iloc[:, 1]
#### test integer
ser.iloc[0] = 9
assert ser.values == [9, 1]
with pytest.raises(IndexError):
ser.iloc[6] = 100
#### test slice
ser.iloc[:] = [20, 30]
assert ser.values == [20, 30]
ser.iloc[:300] = 3
assert ser.values == [3, 3]
#### test iterable
ser.iloc[[0, 1]] = 100
assert ser.values == [100, 100]
#### test bool
ser.iloc[[True, False]] = [99]
assert ser.values == [99, 100]
# test on a slice of series
ser = pam.Series([0, 1, 2, 3, 4], index=["zero", "one", "two", "three", "four"])
ser2 = ser.iloc[:-1]
with pytest.raises(IndexError):
ser2.iloc[4] = 10
# test loc
ser = pam.Series([0, 1, 2, 3, 4])
ser.loc[0, 1] = 100
assert ser.values == [100, 100, 2, 3, 4]
ser = pam.Series([0, 1, 2, 3, 4])
ser.loc["dne"] = 100
assert ser.index == (0, 1, 2, 3, 4, "dne")
assert ser.values == [0, 1, 2, 3, 4, 100]
ser = pam.Series([0, 1, 2, 3, 4])
ser.loc[1:3] = 100
assert ser.index == (0, 1, 2, 3, 4)
assert ser.values == [0, 100, 100, 3, 4]
with pytest.raises(KeyError):
ser = pam.Series([0, 1, 2, 3, 4])
ser.loc["dne", "dne2"] = 100
with pytest.raises(KeyError):
ser = pam.Series([0, 1, 2, 3, 4])
ser.loc["dne":"dne2"] = 100
def test_ser_getitem():
# Test iloc
## test from series
ser = pam.Series([0, 1, 2, 3, 4], index=["zero", "one", "two", "three", "four"])
#### test integer
assert ser.iloc[0] == 0
assert ser.iloc[4] == 4
with pytest.raises(IndexError):
assert ser.iloc[6] == 100
#### test slice
assert ser.iloc[:].values == [0, 1, 2, 3, 4]
assert ser.iloc[:3].values == [0, 1, 2]
assert ser.iloc[:10].values == [0, 1, 2, 3, 4]
#### test iterable
assert ser.iloc[[0, 1, 2]].values == [0, 1, 2]
#### test bool
assert ser.iloc[[True, True, False, False, True]].values == [0, 1, 4]
assert ser.iloc[pam.Series([True, True, False, False, True])].values == [0, 1, 4]
# test from series view of a dataframe
df = pam.DataFrame(
[[11, 12, 13, 14, 15], [0, 1, 2, 3, 4]],
columns=["zero", "one", "two", "three", "four"],
)
ser = df.iloc[1, :]
#### test integer
assert ser.iloc[0] == 0
assert ser.iloc[4] == 4
with pytest.raises(IndexError):
assert ser.iloc[6] == 100
#### test slice
assert ser.iloc[:].values == [0, 1, 2, 3, 4]
assert ser.iloc[:3].values == [0, 1, 2]
assert ser.iloc[:10].values == [0, 1, 2, 3, 4]
#### test iterable
assert ser.iloc[[0, 1, 2]].values == [0, 1, 2]
#### test bool
assert ser.iloc[[True, True, False, False, True]].values == [0, 1, 4]
assert ser.iloc[pam.Series([True, True, False, False, True])].values == [0, 1, 4]
# test on a slice of series
ser = pam.Series([0, 1, 2, 3, 4], index=["zero", "one", "two", "three", "four"])
ser2 = ser.iloc[:-1]
assert ser2.iloc[0:6].values == [0, 1, 2, 3]
with pytest.raises(IndexError):
assert ser2.iloc[5] == 4
# test loc
## test tuple
ser = pam.Series([0, 1, 2, 3], index=[3, 2, 1, 0])
assert ser.loc[(0, 1, 2, 3)] == [3, 2, 1, 0]
# test ser getitem
ser = pam.Series([0, 1, 2, 3], index=[3, 2, 1, 0])
with pytest.raises(KeyError):
assert ser[0, 1, 2, 4] == [3, 2, 1, 0]
with pytest.raises(KeyError):
assert ser[4] == nan
assert ser[[0, 1, 2, 3]] == [3, 2, 1, 0]
assert ser[[0, 1, 100]] == [3, 2, nan]
def test_concat():
# test with Series
ser1 = pam.Series([1, 2, 3], name="1", index=["one", "two", "three"])
ser2 = pam.Series([1, 3, 4], name="2", index=["one", "three", "four"])
assert pam.concat([ser1, ser2]) == pam.Series(
[1, 2, 3, 1, 3, 4], index=["one", "two", "three", "one", "three", "four"]
)
assert pam.concat([ser1, ser2], axis=1).equals(
pam.DataFrame(
{"1": [1, 2, 3, nan], "2": [1, nan, 3, 4]},
index=["one", "two", "three", "four"],
)
)
assert pam.concat([ser1, ser2], axis=1, join="inner").equals(
pam.DataFrame(
{"1": [1, 3], "2": [1, 3]},
index=[
"one",
"three",
],
)
)
assert pam.concat([ser1, ser2], axis=1, ignore_index=True).equals(
pam.DataFrame(
{0: [1, 2, 3, nan], 1: [1, nan, 3, 4]},
index=["one", "two", "three", "four"],
)
)
# Test with DataFrames
df1 = pam.DataFrame(
[["a", 1, "a"], ["b", 2, "b"]], columns=["letter", "number", "date"]
)
df2 = pam.DataFrame([["c", 3], ["d", 4]], columns=["letter", "number"])
assert pam.concat([df1, df2]).equals(
pam.DataFrame(
{
"letter": ["a", "b", "c", "d"],
"number": [1, 2, 3, 4],
"date": ["a", "b", nan, nan],
}
)
)
assert pam.concat([df1, df2], join="inner").equals(
pam.DataFrame({"letter": ["a", "b", "c", "d"], "number": [1, 2, 3, 4]})
)
assert pam.concat([df1, df2], ignore_index=True).equals(
pam.DataFrame(
{
"letter": ["a", "b", "c", "d"],
"number": [1, 2, 3, 4],
"date": ["a", "b", nan, nan],
},
index=[0, 1, 0, 1],
)
)
assert pam.concat([df1, df2], axis=1).equals(
pam.DataFrame(
[["a", 1, "a", "c", 3], ["b", 2, "b", "d", 4]],
index=[0, 1],
columns=["letter", "number", "date", "letter", "number"],
)
)
def test_df_setitem():
# test non-view dataframe
df = pam.DataFrame({"one": [0, 1, 2], "two": [1, 2, 3]})
df.iloc[0, 0] = 99
assert df.iloc[0, 0] == 99
df.iloc[:, 0] = 100
assert df.iloc[:, 0].values == [100, 100, 100]
df.iloc[0, :] = 99
assert df.iloc[0, :].values == [99, 99]
df.iloc[[0, 1, 2], 1] = 9
assert df.iloc[0, :].values == [99, 9]
df.iloc[0:2, 0:2] = [[10, 11], [10, 11]]
assert df.iloc[0:2, 0:2].values == [[10, 11], [10, 11]]
# test single item assignments to multiple cells by slices
df = pam.DataFrame(
{"one": [0, 0, 0, 0], "two": [0, 0, 0, 0], "three": [0, 0, 0, 0]}
)
df.iloc[1:3, 1:3] = 2
assert df.values == [[0, 0, 0], [0, 2, 2], [0, 2, 2], [0, 0, 0]]
df = pam.DataFrame(
{"one": [0, 0, 0, 0], "two": [0, 0, 0, 0], "three": [0, 0, 0, 0]}
)
df.iloc[1:6, 1:6] = 2
assert df.values == [[0, 0, 0], [0, 2, 2], [0, 2, 2], [0, 2, 2]]
# test on view dataframe
df = pam.DataFrame(
{
"nil": [0, 0, 0, 0, 0],
"one": [0, 0, 1, 2, 0],
"two": [0, 1, 2, 3, 0],
"three": [0, 0, 0, 0, 0],
}
)
df1 = df.iloc[1:-1, 1:-1]
df1.iloc[0, 0] = 99
assert df1.iloc[0, 0] == 99
df1.iloc[:, 0] = 100
assert df1.iloc[:, 0].values == [100, 100, 100]
df1.iloc[0, :] = 99
assert df1.iloc[0, :] == [99, 99]
df1.iloc[[0, 1, 2], 1] = 9
assert df1.iloc[0, :] == [9, 9, 9]
df1.iloc[1:50, 1:50] = 999
assert df.values == [
[0, 0, 0, 0],
[0, 99, 9, 0],
[0, 100, 999, 0],
[0, 100, 999, 0],
[0, 0, 0, 0],
]
# Test Boolean setitem
# DataFrame
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df[df > 20] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df.loc[df > 20] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df.iloc[df > 20] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
# 2d boolean array
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df[(df > 20).values] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df.loc[(df > 20).values] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
df = pam.DataFrame([[1, 10, 100, 1000], [2, 20, 200, 2000], [3, 30, 300, 3000]])
df.iloc[(df > 20).values] = 99
assert df.values == [[1, 10, 99, 99], [2, 20, 99, 99], [3, 99, 99, 99]]
# test setitem with an non matching series
df = pam.DataFrame({"one": [10, 10, 10], "two": [20, 20, 20]})
ser = pam.Series([4, 2, 1], index=(2, 1, 0))
df["three"] = ser
assert df["three"].values == [1, 2, 4]
# test setitem with an non matching series
df = pam.DataFrame({"one": [10, 10, 10], "two": [20, 20, 20]})
ser = pam.Series([4, 2, 1], index=(3, 1, 0))
df["three"] = ser
assert df["three"].values == [1, 2, nan]
# test rowwise setitem with a non matching index
df = pam.DataFrame({"one": [10, 10, 10], "two": [20, 20, 20]})
ser = pam.Series([2, 1], index=("two", "three"))
df.loc["hello", :] = ser
assert df.loc["hello", :].values == [nan, 2]
import datetime
import pambdas as pam
print("Create DataFrame")
df = pam.DataFrame([[1, "a", 2.0], [2, "b", 3.0]],
columns=["int", "char", "float"])
df = pam.DataFrame({"int": [1, 2], "char": ["a", "b"], "float": [1.0, 2.0]})
print(df)
print("*" * 80)
print("Assign values or apply")
df["int"] = df["int"].apply(lambda x: x * 10)
print(df)
print("*" * 80)
print("Access and set items with loc and iloc")
ser = df["int"]
ser = df.loc[:, "int"]
ser = df.iloc[:, 0]
print(ser)
print("*" * 80)
print("Setting an item on a slice will change the DataFrame")
ser.iloc[0] = 1000
print(df)
print("*" * 80)
print("Append a new series, or do unary operations by Python's rules")
df["result"] = df["int"] + df["float"]
print(df)
print("*" * 80)
