Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

feat: Allow float in interpolate_by by column #18015

Merged
merged 11 commits into from
Aug 18, 2024
Merged

feat: Allow float in interpolate_by by column #18015

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
19e3386
allow float in interpolate_by by column
agossard Aug 2, 2024
f13f211
Fix formatting problems
agossard Aug 2, 2024
f226b98
type hint
agossard Aug 2, 2024
200d098
more formatting
agossard Aug 3, 2024
0e3b77d
less wastefull hypothesis testing parameters
agossard Aug 3, 2024
04c7c2b
formatting
agossard Aug 3, 2024
5b25b45
really?
agossard Aug 3, 2024
9210a08
At float version of test_interpolate_by_trailing_nulls
agossard Aug 3, 2024
bf3af7a
more formatting
agossard Aug 3, 2024
c7c9ca3
use @given instead of parametrize
agossard Aug 5, 2024
c9ff1a8
try float bounds on hypothesis test
agossard Aug 9, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 16 additions & 4 deletions crates/polars-ops/src/series/ops/interpolation/interpolate_by.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,7 @@ fn interpolate_impl_by_sorted<T, F, I>(
) -> PolarsResult<ChunkedArray<T>>
where
T: PolarsNumericType,
F: PolarsIntegerType,
F: PolarsNumericType,
I: Fn(T::Native, T::Native, &[F::Native], &mut Vec<T::Native>),
{
// This implementation differs from pandas as that boundary None's are not removed.
Expand Down Expand Up @@ -169,7 +169,7 @@ fn interpolate_impl_by<T, F, I>(
) -> PolarsResult<ChunkedArray<T>>
where
T: PolarsNumericType,
F: PolarsIntegerType,
F: PolarsNumericType,
I: Fn(T::Native, T::Native, &[F::Native], &mut [T::Native], &[IdxSize]),
{
// This implementation differs from pandas as that boundary None's are not removed.
Expand Down Expand Up @@ -273,7 +273,7 @@ pub fn interpolate_by(s: &Series, by: &Series, by_is_sorted: bool) -> PolarsResu
) -> PolarsResult<Series>
where
T: PolarsNumericType,
F: PolarsIntegerType,
F: PolarsNumericType,
ChunkedArray<T>: IntoSeries,
{
if is_sorted {
Expand All @@ -290,6 +290,18 @@ pub fn interpolate_by(s: &Series, by: &Series, by_is_sorted: bool) -> PolarsResu
}

match (s.dtype(), by.dtype()) {
(DataType::Float64, DataType::Float64) => {
func(s.f64().unwrap(), by.f64().unwrap(), by_is_sorted)
},
(DataType::Float64, DataType::Float32) => {
func(s.f64().unwrap(), by.f32().unwrap(), by_is_sorted)
},
(DataType::Float32, DataType::Float64) => {
func(s.f32().unwrap(), by.f64().unwrap(), by_is_sorted)
},
(DataType::Float32, DataType::Float32) => {
func(s.f32().unwrap(), by.f32().unwrap(), by_is_sorted)
},
(DataType::Float64, DataType::Int64) => {
func(s.f64().unwrap(), by.i64().unwrap(), by_is_sorted)
},
Expand Down Expand Up @@ -326,7 +338,7 @@ pub fn interpolate_by(s: &Series, by: &Series, by_is_sorted: bool) -> PolarsResu
_ => {
polars_bail!(InvalidOperation: "expected series to be Float64, Float32, \
Int64, Int32, UInt64, UInt32, and `by` to be Date, Datetime, Int64, Int32, \
UInt64, or UInt32")
UInt64, UInt32, Float32 or Float64")
},
}
}
23 changes: 19 additions & 4 deletions py-polars/tests/unit/operations/test_interpolate_by.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,8 @@
pl.Int32,
pl.UInt64,
pl.UInt32,
pl.Float32,
pl.Float64
],
)
@pytest.mark.parametrize(
Expand Down Expand Up @@ -143,14 +145,16 @@ def test_interpolate_by_trailing_nulls() -> None:


@given(data=st.data())
def test_interpolate_vs_numpy(data: st.DataObject) -> None:
@pytest.mark.parametrize("x_dtype", [pl.Date, pl.Float64])
Copy link
Collaborator

@MarcoGorelli MarcoGorelli Aug 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this can be in given, you can use st.sampled_from

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

as in, x_dtype=st.sampled_from([pl.Date, pl.Float64])

def test_interpolate_vs_numpy(data: st.DataObject, x_dtype: pl.DataType) -> None:

dataframe = (
data.draw(
dataframes(
[
column(
"ts",
dtype=pl.Date,
dtype=x_dtype,
allow_null=False,
),
column(
Expand All @@ -166,13 +170,24 @@ def test_interpolate_vs_numpy(data: st.DataObject) -> None:
.fill_nan(None)
.unique("ts")
)

if x_dtype == pl.Float64:
assume(not dataframe['ts'].is_nan().any())
assume(not dataframe['ts'].is_null().any())
assume(not dataframe["ts"].is_in([float("-inf"), float("inf")]).any())

assume(not dataframe["value"].is_null().all())
assume(not dataframe["value"].is_in([float("-inf"), float("inf")]).any())

dataframe = dataframe.sort('ts')

result = dataframe.select(pl.col("value").interpolate_by("ts"))["value"]

mask = dataframe["value"].is_not_null()
x = dataframe["ts"].to_numpy().astype("int64")
xp = dataframe["ts"].filter(mask).to_numpy().astype("int64")

np_dtype = "int64" if x_dtype == pl.Date else 'float64'
x = dataframe["ts"].to_numpy().astype(np_dtype)
xp = dataframe["ts"].filter(mask).to_numpy().astype(np_dtype)
yp = dataframe["value"].filter(mask).to_numpy().astype("float64")
interp = np.interp(x, xp, yp)
# Polars preserves nulls on boundaries, but NumPy doesn't.
Expand Down
Loading