def test_arrange_column_spec_inverse2():
actual = dp(mtcars).select(["hp", "qsec"]).arrange(["hp", "qsec"]).pd
should = mtcars.sort_values(["hp", "qsec"],
ascending=[False, False])[["hp", "qsec"]]
assert_frame_equal(should, actual)
actual = dp(mtcars).select(["hp", "qsec"]).arrange(["-hp", "qsec"]).pd
should = mtcars.sort_values(["hp", "qsec"],
ascending=[True, False])[["hp", "qsec"]]
assert_frame_equal(should, actual)
actual = dp(mtcars).select(["hp", "qsec"]).arrange(["hp", "-qsec"]).pd
should = mtcars.sort_values(["hp", "qsec"],
ascending=[False, True])[["hp", "qsec"]]
assert_frame_equal(should, actual)
def test_summary_quantiles():
args = [("disp", lambda x, q=q: x.quantile(q), "q%.2f" % q)
for q in np.arange(0, 1.1, 0.1)]
actual = dp(mtcars).sort_values("cyl").groupby("cyl").summarise(*args).pd
lambdas = [lambda x, q=q: x.quantile(q) for q in np.arange(0, 1.1, 0.1)]
for l, q in zip(lambdas, np.arange(0, 1.1, 0.1)):
l.__name__ = "q%.2f" % q
should = (mtcars.sort_values("cyl").groupby("cyl")["disp"].aggregate(
lambdas).reset_index())
assert_frame_equal(should, actual)
def test_mixing_context_manager_and_dp():
with dppd(mtcars) as (dpY, Y):
dpY.sort_values("hp")
dp(diamonds).filter_by(X.cut == "ideal")
dpY.filter_by(Y.cyl.isin([4, 6]))
actual_diamonds = dp().sort_values("price").head().pd
actual_mtcars_full = dpY.pd
dpY.head()
actual_mtcars = dpY.pd
should_diamonds = diamonds[diamonds.cut == "ideal"].sort_values(
"price").head()
should_mtcars = mtcars.sort_values("hp")
should_mtcars_full = should_mtcars[should_mtcars["cyl"].isin([4, 6])]
should_mtcars = should_mtcars_full.head()
assert_frame_equal(should_diamonds, actual_diamonds)
assert_frame_equal(should_mtcars, actual_mtcars)
assert_frame_equal(should_mtcars_full, actual_mtcars_full)
def test_arrange_column_spec_inverse():
actual = dp(mtcars).select("hp").arrange("-hp").pd
should = mtcars.sort_values("hp", ascending=True)[["hp"]]
assert_frame_equal(should, actual)
def test_sorting_within_groups():
actual = dp(mtcars).groupby(X.cyl).arrange("qsec").ungroup().pd
should = mtcars.sort_values(["cyl", "qsec"])
should = should[actual.columns]
assert_frame_equal(should, actual)