dycw-utilities 0.109.20__tar.gz → 0.109.22__tar.gz

{dycw_utilities-0.109.20 → dycw_utilities-0.109.22}/PKG-INFO

@@ -1,12 +1,12 @@
 Metadata-Version: 2.4
 Name: dycw-utilities
-Version: 0.109.20
+Version: 0.109.22
 Author-email: Derek Wan <d.wan@icloud.com>
 License-File: LICENSE
 Requires-Python: >=3.12
 Requires-Dist: typing-extensions<4.14,>=4.13.1
 Provides-Extra: test
-Requires-Dist: hypothesis<6.132,>=6.131.7; extra == 'test'
+Requires-Dist: hypothesis<6.132,>=6.131.9; extra == 'test'
 Requires-Dist: pytest-asyncio<0.27,>=0.26.0; extra == 'test'
 Requires-Dist: pytest-cov<6.2,>=6.1.1; extra == 'test'
 Requires-Dist: pytest-instafail<0.6,>=0.5.0; extra == 'test'
@@ -80,7 +80,7 @@ Provides-Extra: zzz-test-hypothesis
 Requires-Dist: aiosqlite<0.22,>=0.21.0; extra == 'zzz-test-hypothesis'
 Requires-Dist: asyncpg<0.31,>=0.30.0; extra == 'zzz-test-hypothesis'
 Requires-Dist: greenlet<3.3,>=3.2.0; extra == 'zzz-test-hypothesis'
-Requires-Dist: hypothesis<6.132,>=6.131.7; extra == 'zzz-test-hypothesis'
+Requires-Dist: hypothesis<6.132,>=6.131.9; extra == 'zzz-test-hypothesis'
 Requires-Dist: luigi<3.7,>=3.6.0; extra == 'zzz-test-hypothesis'
 Requires-Dist: numpy<2.3,>=2.2.5; extra == 'zzz-test-hypothesis'
 Requires-Dist: pathvalidate<3.3,>=3.2.3; extra == 'zzz-test-hypothesis'

{dycw_utilities-0.109.20 → dycw_utilities-0.109.22}/pyproject.toml

@@ -27,7 +27,7 @@ dev = [
   "fpdf2 >= 2.8.3, < 2.9",
   "greenlet >= 3.2.0, < 3.3",  # for sqlalchemy async
   "httpx >= 0.28.1, < 0.29",  # for fastapi
-  "hypothesis >= 6.131.7, < 6.132",
+  "hypothesis >= 6.131.9, < 6.132",
   "img2pdf >= 0.6.0, < 0.7",
   "lightweight-charts >= 2.1, < 2.2",
   "loguru >= 0.7.3, < 0.8",
@@ -56,6 +56,7 @@ dev = [
   "scipy >= 1.15.2, < 1.16",
   "slack-sdk >= 3.35.0, < 3.36",
   "sqlalchemy >= 2.0.40, < 2.1",
+  "statsmodels >= 0.14.4, < 0.15",
   "streamlit >= 1.44.1, < 1.45",
   "tenacity >= 8.5.0, < 9.0",  # limited by luigi
   "tomlkit >= 0.13.2, < 0.14",
@@ -92,11 +93,11 @@ dependencies = [
 name = "dycw-utilities"
 readme = "README.md"
 requires-python = ">= 3.12"
-version = "0.109.20"
+version = "0.109.22"
 
 [project.optional-dependencies]
 test = [
-  "hypothesis >= 6.131.7, < 6.132",
+  "hypothesis >= 6.131.9, < 6.132",
   "pytest >= 8.3.5, < 8.4",
   "pytest-asyncio >= 0.26.0, < 0.27",
   "pytest-cov >= 6.1.1, < 6.2",
@@ -174,7 +175,7 @@ zzz-test-hypothesis = [
   "aiosqlite >= 0.21.0, < 0.22",
   "asyncpg >= 0.30.0, < 0.31",  # for sqlalchemy async
   "greenlet >= 3.2.0, < 3.3",  # for sqlalchemy async
-  "hypothesis >= 6.131.7, < 6.132",
+  "hypothesis >= 6.131.9, < 6.132",
   "luigi >= 3.6.0, < 3.7",
   "numpy >= 2.2.5, < 2.3",
   "pathvalidate >= 3.2.3, < 3.3",
@@ -335,7 +336,7 @@ zzz-test-zoneinfo = [
 # bump-my-version
 [tool.bumpversion]
 allow_dirty = true
-current_version = "0.109.20"
+current_version = "0.109.22"
 
 [[tool.bumpversion.files]]
 filename = "src/utilities/__init__.py"

{dycw_utilities-0.109.20 → dycw_utilities-0.109.22}/src/tests/test_polars.py

@@ -43,6 +43,7 @@ from polars import (
     Series,
     String,
     Struct,
+    UInt32,
     col,
     date_range,
     datetime_range,
@@ -107,6 +108,8 @@ from utilities.polars import (
     _InsertBetweenMissingColumnsError,
     _InsertBetweenNonConsecutiveError,
     _yield_struct_series_element_remove_nulls,
+    ac_halflife,
+    acf,
     adjust_frequencies,
     append_dataclass,
     are_frames_equal,
@@ -168,6 +171,66 @@ if TYPE_CHECKING:
     from utilities.types import MaybeType, StrMapping, WeekDay
 
 
+class TestACF:
+    def test_main(self) -> None:
+        series = Series(linspace(0, 2 * pi, 1000))
+        df = acf(series)
+        check_polars_dataframe(
+            df, height=31, schema_list={"lag": UInt32, "autocorrelation": Float64}
+        )
+
+    def test_alpha(self) -> None:
+        series = Series(linspace(0, 2 * pi, 1000))
+        df = acf(series, alpha=0.5)
+        check_polars_dataframe(
+            df,
+            height=31,
+            schema_list={
+                "lag": UInt32,
+                "autocorrelation": Float64,
+                "lower": Float64,
+                "upper": Float64,
+            },
+        )
+
+    def test_qstat(self) -> None:
+        series = Series(linspace(0, 2 * pi, 1000))
+        df = acf(series, qstat=True)
+        check_polars_dataframe(
+            df,
+            height=31,
+            schema_list={
+                "lag": UInt32,
+                "autocorrelation": Float64,
+                "qstat": Float64,
+                "pvalue": Float64,
+            },
+        )
+
+    def test_alpha_and_qstat(self) -> None:
+        series = Series(linspace(0, 2 * pi, 1000))
+        df = acf(series, alpha=0.5, qstat=True)
+        check_polars_dataframe(
+            df,
+            height=31,
+            schema_list={
+                "lag": UInt32,
+                "autocorrelation": Float64,
+                "lower": Float64,
+                "upper": Float64,
+                "qstat": Float64,
+                "pvalue": Float64,
+            },
+        )
+
+
+class TestACHalfLife:
+    def test_main(self) -> None:
+        series = Series(linspace(0, 2 * pi, 1000))
+        halflife = ac_halflife(series)
+        assert halflife == 169.94
+
+
 class TestAdjustFrequencies:
     def test_main(self) -> None:
         n = 1000

dycw_utilities-0.109.22/src/tests/test_statsmodel.py

@@ -0,0 +1,51 @@
+from __future__ import annotations
+
+from numpy import allclose, isclose, linspace, pi
+
+from utilities.statsmodels import ac_halflife, acf
+
+
+class TestACF:
+    def test_main(self) -> None:
+        x = linspace(0, 2 * pi, 1000)
+        acfs = acf(x)
+        assert acfs.shape == (31,)
+        assert isclose(acfs[-1], 0.9100539400539398)
+
+    def test_alpha(self) -> None:
+        x = linspace(0, 2 * pi, 1000)
+        result = acf(x, alpha=0.5)
+        acfs, confint = result
+        assert acfs.shape == (31,)
+        assert confint.shape == (31, 2)
+        assert allclose(confint[-1, :], [0.7534104755915192, 1.0666974045163604])
+
+    def test_qstat(self) -> None:
+        x = linspace(0, 2 * pi, 1000)
+        result = acf(x, qstat=True)
+        acfs, qstat, pvalues = result
+        assert acfs.shape == (31,)
+        assert qstat.shape == (30,)
+        assert isclose(qstat[-1], 27769.955439016478)
+        assert pvalues.shape == (30,)
+        assert isclose(pvalues[-1], 0.0)
+
+    def test_alpha_and_qstat(self) -> None:
+        x = linspace(0, 2 * pi, 1000)
+        result = acf(x, alpha=0.5, qstat=True)
+        acfs, confint, qstat, pvalues = result
+        assert acfs.shape == (31,)
+        assert isclose(acfs[-1], 0.9100539400539398)
+        assert confint.shape == (31, 2)
+        assert allclose(confint[-1, :], [0.7534104755915192, 1.0666974045163604])
+        assert qstat.shape == (30,)
+        assert isclose(qstat[-1], 27769.955439016478)
+        assert pvalues.shape == (30,)
+        assert isclose(pvalues[-1], 0.0)
+
+
+class TestACHalfLife:
+    def test_main(self) -> None:
+        x = linspace(0, 2 * pi, 1000)
+        halflife = ac_halflife(x)
+        assert halflife == 169.94

{dycw_utilities-0.109.20 → dycw_utilities-0.109.22}/src/utilities/__init__.py

@@ -1,3 +1,3 @@
 from __future__ import annotations
 
-__version__ = "0.109.20"
+__version__ = "0.109.22"

{dycw_utilities-0.109.20 → dycw_utilities-0.109.22}/src/utilities/polars.py

@@ -121,6 +121,7 @@ if TYPE_CHECKING:
     )
 
     from utilities.numpy import NDArrayB, NDArrayF
+    from utilities.statsmodels import ACFMissing
     from utilities.types import (
         Dataclass,
         MaybeIterable,
@@ -139,6 +140,116 @@ DatetimeUSEastern = Datetime(time_zone="US/Eastern")
 DatetimeUTC = Datetime(time_zone="UTC")
 _FINITE_EWM_MIN_WEIGHT = 0.9999
 
+##
+
+
+def ac_halflife(
+    series: Series,
+    /,
+    *,
+    adjusted: bool = False,
+    fft: bool = True,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+    step: float = 0.01,
+) -> float:
+    """Compute the autocorrelation halflife."""
+    import utilities.statsmodels
+
+    array = series.to_numpy()
+    return utilities.statsmodels.ac_halflife(
+        array,
+        adjusted=adjusted,
+        fft=fft,
+        bartlett_confint=bartlett_confint,
+        missing=missing,
+        step=step,
+    )
+
+
+##
+
+
+def acf(
+    series: Series,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: bool = False,
+    fft: bool = True,
+    alpha: float | None = None,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> DataFrame:
+    """Compute the autocorrelations of a series."""
+    from numpy import ndarray
+
+    import utilities.statsmodels
+
+    array = series.to_numpy()
+    result = utilities.statsmodels.acf(
+        array,
+        adjusted=adjusted,
+        nlags=nlags,
+        qstat=qstat,
+        fft=fft,
+        alpha=alpha,
+        bartlett_confint=bartlett_confint,
+        missing=missing,
+    )
+    match result:
+        case ndarray() as acfs:
+            return _acf_process_acfs(acfs)
+        case ndarray() as acfs, ndarray() as confints:
+            df_acfs = _acf_process_acfs(acfs)
+            df_confints = _acf_process_confints(confints)
+            return df_acfs.join(df_confints, on=["lag"])
+        case ndarray() as acfs, ndarray() as qstats, ndarray() as pvalues:
+            df_acfs = _acf_process_acfs(acfs)
+            df_qstats_pvalues = _acf_process_qstats_pvalues(qstats, pvalues)
+            return df_acfs.join(df_qstats_pvalues, on=["lag"], how="left")
+        case (
+            ndarray() as acfs,
+            ndarray() as confints,
+            ndarray() as qstats,
+            ndarray() as pvalues,
+        ):
+            df_acfs = _acf_process_acfs(acfs)
+            df_confints = _acf_process_confints(confints)
+            df_qstats_pvalues = _acf_process_qstats_pvalues(qstats, pvalues)
+            return join(df_acfs, df_confints, df_qstats_pvalues, on=["lag"], how="left")
+        case _ as never:
+            assert_never(never)
+
+
+def _acf_process_acfs(acfs: NDArrayF, /) -> DataFrame:
+    return (
+        Series(name="autocorrelation", values=acfs, dtype=Float64)
+        .to_frame()
+        .with_row_index(name="lag")
+    )
+
+
+def _acf_process_confints(confints: NDArrayF, /) -> DataFrame:
+    return DataFrame(
+        data=confints, schema={"lower": Float64, "upper": Float64}
+    ).with_row_index(name="lag")
+
+
+def _acf_process_qstats_pvalues(qstats: NDArrayF, pvalues: NDArrayF, /) -> DataFrame:
+    from numpy import hstack
+
+    data = hstack([qstats.reshape(-1, 1), pvalues.reshape(-1, 1)])
+    return DataFrame(
+        data=data, schema={"qstat": Float64, "pvalue": Float64}
+    ).with_row_index(name="lag", offset=1)
+
+
+##
+
+
+# def acf_halflife(series: Series,/)
 
 ##
 
@@ -1840,6 +1951,7 @@ __all__ = [
     "SetFirstRowAsColumnsError",
     "StructFromDataClassError",
     "YieldStructSeriesElementsError",
+    "acf",
     "adjust_frequencies",
     "append_dataclass",
     "are_frames_equal",

dycw_utilities-0.109.22/src/utilities/statsmodels.py

@@ -0,0 +1,150 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any, Literal, cast, overload
+
+import numpy as np
+import statsmodels.tsa.stattools
+from numpy import arange, argmax, interp, nan
+
+from utilities.numpy import shift
+
+if TYPE_CHECKING:
+    from utilities.numpy import NDArrayF
+
+
+def ac_halflife(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    fft: bool = True,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+    step: float = 0.01,
+) -> float:
+    """Compute the autocorrelation halflife."""
+    (n,) = array.shape
+    acfs = acf(
+        array,
+        adjusted=adjusted,
+        nlags=n,
+        fft=fft,
+        bartlett_confint=bartlett_confint,
+        missing=missing,
+    )
+    lags = arange(0, n, step=step)
+    interp_acfs = interp(lags, arange(n), acfs)
+    is_half = (shift(interp_acfs) > 0.5) & (interp_acfs <= 0.5)
+    return lags[argmax(is_half)].item() if np.any(is_half) else nan
+
+
+##
+
+
+type ACFMissing = Literal["none", "raise", "conservative", "drop"]
+
+
+@overload
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: Literal[False] = False,
+    fft: bool = True,
+    alpha: None = None,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> NDArrayF: ...
+@overload
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: Literal[False] = False,
+    fft: bool = True,
+    alpha: float,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> tuple[NDArrayF, NDArrayF]: ...
+@overload
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: Literal[True],
+    fft: bool = True,
+    alpha: float,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> tuple[NDArrayF, NDArrayF, NDArrayF, NDArrayF]: ...
+@overload
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: Literal[True],
+    fft: bool = True,
+    alpha: None = None,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> tuple[NDArrayF, NDArrayF, NDArrayF]: ...
+@overload
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: bool = False,
+    fft: bool = True,
+    alpha: float | None = None,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> (
+    NDArrayF
+    | tuple[NDArrayF, NDArrayF]
+    | tuple[NDArrayF, NDArrayF, NDArrayF]
+    | tuple[NDArrayF, NDArrayF, NDArrayF, NDArrayF]
+): ...
+def acf(
+    array: NDArrayF,
+    /,
+    *,
+    adjusted: bool = False,
+    nlags: int | None = None,
+    qstat: bool = False,
+    fft: bool = True,
+    alpha: float | None = None,
+    bartlett_confint: bool = True,
+    missing: ACFMissing = "none",
+) -> (
+    NDArrayF
+    | tuple[NDArrayF, NDArrayF]
+    | tuple[NDArrayF, NDArrayF, NDArrayF]
+    | tuple[NDArrayF, NDArrayF, NDArrayF, NDArrayF]
+):
+    """Typed version of `acf`."""
+    return cast(
+        "Any",
+        statsmodels.tsa.stattools.acf(
+            array,
+            adjusted=adjusted,
+            nlags=nlags,
+            qstat=qstat,
+            fft=fft,
+            alpha=alpha,
+            bartlett_confint=bartlett_confint,
+            missing=missing,
+        ),
+    )
+
+
+__all__ = ["ACFMissing", "ac_halflife", "acf"]