scipy 1.15.3__cp312-cp312-macosx_14_0_arm64.whl → 1.16.0rc2__cp312-cp312-macosx_14_0_arm64.whl

scipy/stats/_morestats.py

@@ -15,15 +15,15 @@ from scipy._lib._util import _rename_parameter, _contains_nan, _get_nan
 from scipy._lib._array_api import (
     array_namespace,
     xp_size,
-    xp_moveaxis_to_end,
     xp_vector_norm,
+    xp_promote,
 )
 
 from ._ansari_swilk_statistics import gscale, swilk
 from . import _stats_py, _wilcoxon
 from ._fit import FitResult
 from ._stats_py import (find_repeats, _get_pvalue, SignificanceResult,  # noqa:F401
-                        _SimpleNormal, _SimpleChi2)
+                        _SimpleNormal, _SimpleChi2, _length_nonmasked)
 from .contingency import chi2_contingency
 from . import distributions
 from ._distn_infrastructure import rv_generic
@@ -222,7 +222,7 @@ def mvsdist(data):
 
 
 @_axis_nan_policy_factory(
-    lambda x: x, result_to_tuple=lambda x: (x,), n_outputs=1, default_axis=None
+    lambda x: x, result_to_tuple=lambda x, _: (x,), n_outputs=1, default_axis=None
 )
 def kstat(data, n=2, *, axis=None):
     r"""
@@ -309,7 +309,7 @@ def kstat(data, n=2, *, axis=None):
         data = xp.reshape(data, (-1,))
         axis = 0
 
-    N = data.shape[axis]
+    N = _length_nonmasked(data, axis, xp=xp)
 
     S = [None] + [xp.sum(data**k, axis=axis) for k in range(1, n + 1)]
     if n == 1:
@@ -327,7 +327,7 @@ def kstat(data, n=2, *, axis=None):
 
 
 @_axis_nan_policy_factory(
-    lambda x: x, result_to_tuple=lambda x: (x,), n_outputs=1, default_axis=None
+    lambda x: x, result_to_tuple=lambda x, _: (x,), n_outputs=1, default_axis=None
 )
 def kstatvar(data, n=2, *, axis=None):
     r"""Return an unbiased estimator of the variance of the k-statistic.
@@ -375,7 +375,7 @@ def kstatvar(data, n=2, *, axis=None):
     if axis is None:
         data = xp.reshape(data, (-1,))
         axis = 0
-    N = data.shape[axis]
+    N = _length_nonmasked(data, axis, xp=xp)
 
     if n == 1:
         return kstat(data, n=2, axis=axis, _no_deco=True) * 1.0/N
@@ -854,26 +854,24 @@ def ppcc_plot(x, a, b, dist='tukeylambda', plot=None, N=80):
     return svals, ppcc
 
 
-def _log_mean(logx):
+def _log_mean(logx, axis):
     # compute log of mean of x from log(x)
-    res = special.logsumexp(logx, axis=0) - math.log(logx.shape[0])
-    return res
+    return (
+        special.logsumexp(logx, axis=axis, keepdims=True)
+        - math.log(logx.shape[axis])
+    )
 
 
-def _log_var(logx, xp):
+def _log_var(logx, xp, axis):
     # compute log of variance of x from log(x)
-    logmean = _log_mean(logx)
-    # get complex dtype with component dtypes same as `logx` dtype;
-    # see data-apis/array-api#841
-    dtype = xp.result_type(logx.dtype, xp.complex64)
-    pij = xp.full(logx.shape, pi * 1j, dtype=dtype)
-    logxmu = special.logsumexp(xp.stack((logx, logmean + pij)), axis=0)
-    res = (xp.real(xp.asarray(special.logsumexp(2 * logxmu, axis=0)))
-           - math.log(logx.shape[0]))
-    return res
+    logmean = xp.broadcast_to(_log_mean(logx, axis=axis), logx.shape)
+    ones = xp.ones_like(logx)
+    logxmu, _ = special.logsumexp(xp.stack((logx, logmean), axis=0), axis=0,
+                                  b=xp.stack((ones, -ones), axis=0), return_sign=True)
+    return special.logsumexp(2 * logxmu, axis=axis) - math.log(logx.shape[axis])
 
 
-def boxcox_llf(lmb, data):
+def boxcox_llf(lmb, data, *, axis=0, keepdims=False, nan_policy='propagate'):
     r"""The boxcox log-likelihood function.
 
     Parameters
@@ -884,6 +882,26 @@ def boxcox_llf(lmb, data):
         Data to calculate Box-Cox log-likelihood for.  If `data` is
         multi-dimensional, the log-likelihood is calculated along the first
         axis.
+    axis : int, default: 0
+        If an int, the axis of the input along which to compute the statistic.
+        The statistic of each axis-slice (e.g. row) of the input will appear in a
+        corresponding element of the output.
+        If ``None``, the input will be raveled before computing the statistic.
+    nan_policy : {'propagate', 'omit', 'raise'
+        Defines how to handle input NaNs.
+
+        - ``propagate``: if a NaN is present in the axis slice (e.g. row) along
+          which the  statistic is computed, the corresponding entry of the output
+          will be NaN.
+        - ``omit``: NaNs will be omitted when performing the calculation.
+          If insufficient data remains in the axis slice along which the
+          statistic is computed, the corresponding entry of the output will be
+          NaN.
+        - ``raise``: if a NaN is present, a ``ValueError`` will be raised.
+    keepdims : bool, default: False
+        If this is set to True, the axes which are reduced are left
+        in the result as dimensions with size one. With this option,
+        the result will broadcast correctly against the input array.
 
     Returns
     -------
@@ -897,14 +915,17 @@ def boxcox_llf(lmb, data):
 
     Notes
     -----
-    The Box-Cox log-likelihood function is defined here as
+    The Box-Cox log-likelihood function :math:`l` is defined here as
 
     .. math::
 
-        llf = (\lambda - 1) \sum_i(\log(x_i)) -
-              N/2 \log(\sum_i (y_i - \bar{y})^2 / N),
+        l = (\lambda - 1) \sum_i^N \log(x_i) -
+              \frac{N}{2} \log\left(\sum_i^N (y_i - \bar{y})^2 / N\right),
 
-    where ``y`` is the Box-Cox transformed input data ``x``.
+    where :math:`N` is the number of data points ``data`` and :math:`y` is the Box-Cox
+    transformed input data.
+    This corresponds to the *profile log-likelihood* of the original data :math:`x`
+    with some constant terms dropped.
 
     Examples
     --------
@@ -953,32 +974,39 @@ def boxcox_llf(lmb, data):
     >>> plt.show()
 
     """
+    # _axis_nan_policy decorator does not currently support these for non-NumPy arrays
+    kwargs = {}
+    if keepdims is not False:
+        kwargs[keepdims] = keepdims
+    if nan_policy != 'propagate':
+        kwargs[nan_policy] = nan_policy
+    return _boxcox_llf(data, lmb=lmb, axis=axis, **kwargs)
+
+
+@_axis_nan_policy_factory(lambda x: x, n_outputs=1, default_axis=0,
+                          result_to_tuple=lambda x, _: (x,))
+def _boxcox_llf(data, axis=0, *, lmb):
     xp = array_namespace(data)
-    data = xp.asarray(data)
-    N = data.shape[0]
+    lmb, data = xp_promote(lmb, data, force_floating=True, xp=xp)
+    N = data.shape[axis]
     if N == 0:
-        return xp.nan
-
-    dt = data.dtype
-    if xp.isdtype(dt, 'integral'):
-        data = xp.asarray(data, dtype=xp.float64)
-        dt = xp.float64
+        return _get_nan(data, xp=xp)
 
     logdata = xp.log(data)
 
     # Compute the variance of the transformed data.
     if lmb == 0:
-        logvar = xp.log(xp.var(logdata, axis=0))
+        logvar = xp.log(xp.var(logdata, axis=axis))
     else:
         # Transform without the constant offset 1/lmb.  The offset does
         # not affect the variance, and the subtraction of the offset can
         # lead to loss of precision.
         # Division by lmb can be factored out to enhance numerical stability.
         logx = lmb * logdata
-        logvar = _log_var(logx, xp) - 2 * math.log(abs(lmb))
+        logvar = _log_var(logx, xp, axis) - 2 * math.log(abs(lmb))
 
-    res = (lmb - 1) * xp.sum(logdata, axis=0) - N/2 * logvar
-    res = xp.astype(res, dt)
+    res = (lmb - 1) * xp.sum(logdata, axis=axis) - N/2 * logvar
+    res = xp.astype(res, data.dtype, copy=False)  # compensate for NumPy <2.0
     res = res[()] if res.ndim == 0 else res
     return res
 
@@ -1082,10 +1110,15 @@ def boxcox(x, lmbda=None, alpha=None, optimizer=None):
 
     Notes
     -----
-    The Box-Cox transform is given by::
+    The Box-Cox transform is given by:
+
+    .. math::
 
-        y = (x**lmbda - 1) / lmbda,  for lmbda != 0
-            log(x),                  for lmbda = 0
+        y =
+        \begin{cases}
+        \frac{x^\lambda - 1}{\lambda}, &\text{for } \lambda \neq 0
+        \log(x),                       &\text{for } \lambda = 0
+        \end{cases}
 
     `boxcox` requires the input data to be positive.  Sometimes a Box-Cox
     transformation provides a shift parameter to achieve this; `boxcox` does
@@ -1097,9 +1130,9 @@ def boxcox(x, lmbda=None, alpha=None, optimizer=None):
 
     .. math::
 
-        llf(\hat{\lambda}) - llf(\lambda) < \frac{1}{2}\chi^2(1 - \alpha, 1),
+        l(\hat{\lambda}) - l(\lambda) < \frac{1}{2}\chi^2(1 - \alpha, 1),
 
-    with ``llf`` the log-likelihood function and :math:`\chi^2` the chi-squared
+    with :math:`l` the log-likelihood function and :math:`\chi^2` the chi-squared
     function.
 
     References
@@ -1538,12 +1571,24 @@ def yeojohnson(x, lmbda=None):
 
     Notes
     -----
-    The Yeo-Johnson transform is given by::
+    The Yeo-Johnson transform is given by:
 
-        y = ((x + 1)**lmbda - 1) / lmbda,                for x >= 0, lmbda != 0
-            log(x + 1),                                  for x >= 0, lmbda = 0
-            -((-x + 1)**(2 - lmbda) - 1) / (2 - lmbda),  for x < 0, lmbda != 2
-            -log(-x + 1),                                for x < 0, lmbda = 2
+    .. math::
+
+        y =
+        \begin{cases}
+        \frac{(x + 1)^\lambda - 1}{\lambda},
+        &\text{for } x \geq 0, \lambda \neq 0
+        \\
+        \log(x + 1),
+        &\text{for } x \geq 0, \lambda = 0
+        \\
+        -\frac{(-x + 1)^{2 - \lambda} - 1}{2 - \lambda},
+        &\text{for } x < 0, \lambda \neq 2
+        \\
+        -\log(-x + 1),
+        &\text{for } x < 0, \lambda = 2
+        \end{cases}
 
     Unlike `boxcox`, `yeojohnson` does not require the input data to be
     positive.
@@ -1651,15 +1696,18 @@ def yeojohnson_llf(lmb, data):
 
     Notes
     -----
-    The Yeo-Johnson log-likelihood function is defined here as
+    The Yeo-Johnson log-likelihood function :math:`l` is defined here as
 
     .. math::
 
-        llf = -N/2 \log(\hat{\sigma}^2) + (\lambda - 1)
-              \sum_i \text{ sign }(x_i)\log(|x_i| + 1)
+        l = -\frac{N}{2} \log(\hat{\sigma}^2) + (\lambda - 1)
+              \sum_i^N \text{sign}(x_i) \log(|x_i| + 1)
 
-    where :math:`\hat{\sigma}^2` is estimated variance of the Yeo-Johnson
-    transformed input data ``x``.
+    where :math:`N` is the number of data points :math:`x`=``data`` and
+    :math:`\hat{\sigma}^2` is the estimated variance of the Yeo-Johnson transformed
+    input data :math:`x`.
+    This corresponds to the *profile log-likelihood* of the original data :math:`x`
+    with some constant terms dropped.
 
     .. versionadded:: 1.2.0
 
@@ -2884,7 +2932,7 @@ def bartlett(*samples, axis=0):
         raise ValueError("Must enter at least two input sample vectors.")
 
     samples = _broadcast_arrays(samples, axis=axis, xp=xp)
-    samples = [xp_moveaxis_to_end(sample, axis, xp=xp) for sample in samples]
+    samples = [xp.moveaxis(sample, axis, -1) for sample in samples]
 
     Ni = [xp.asarray(sample.shape[-1], dtype=sample.dtype) for sample in samples]
     Ni = [xp.broadcast_to(N, samples[0].shape[:-1]) for N in Ni]
@@ -2893,14 +2941,13 @@ def bartlett(*samples, axis=0):
     ssq = [arr[xp.newaxis, ...] for arr in ssq]
     Ni = xp.concat(Ni, axis=0)
     ssq = xp.concat(ssq, axis=0)
-    # sum dtype can be removed when 2023.12 rules kick in
     dtype = Ni.dtype
-    Ntot = xp.sum(Ni, axis=0, dtype=dtype)
+    Ntot = xp.sum(Ni, axis=0)
     spsq = xp.sum((Ni - 1)*ssq, axis=0, dtype=dtype) / (Ntot - k)
     numer = ((Ntot - k) * xp.log(spsq)
              - xp.sum((Ni - 1)*xp.log(ssq), axis=0, dtype=dtype))
     denom = (1 + 1/(3*(k - 1))
-             * ((xp.sum(1/(Ni - 1), axis=0, dtype=dtype)) - 1/(Ntot - k)))
+             * ((xp.sum(1/(Ni - 1), axis=0)) - 1/(Ntot - k)))
     T = numer / denom
 
     chi2 = _SimpleChi2(xp.asarray(k-1))
@@ -3449,7 +3496,7 @@ def mood(x, y, axis=0, alternative="two-sided"):
 WilcoxonResult = _make_tuple_bunch('WilcoxonResult', ['statistic', 'pvalue'])
 
 
-def wilcoxon_result_unpacker(res):
+def wilcoxon_result_unpacker(res, _):
     if hasattr(res, 'zstatistic'):
         return res.statistic, res.pvalue, res.zstatistic
     else:
@@ -3870,16 +3917,15 @@ def median_test(*samples, ties='below', correction=True, lambda_=1,
     # Validate the sizes and shapes of the arguments.
     for k, d in enumerate(data):
         if d.size == 0:
-            raise ValueError("Sample %d is empty. All samples must "
-                             "contain at least one value." % (k + 1))
+            raise ValueError(f"Sample {k + 1} is empty. All samples must "
+                             f"contain at least one value.")
         if d.ndim != 1:
-            raise ValueError("Sample %d has %d dimensions.  All "
-                             "samples must be one-dimensional sequences." %
-                             (k + 1, d.ndim))
+            raise ValueError(f"Sample {k + 1} has {d.ndim} dimensions. "
+                             f"All samples must be one-dimensional sequences.")
 
     cdata = np.concatenate(data)
-    contains_nan, nan_policy = _contains_nan(cdata, nan_policy)
-    if contains_nan and nan_policy == 'propagate':
+    contains_nan = _contains_nan(cdata, nan_policy)
+    if nan_policy == 'propagate' and contains_nan:
         return MedianTestResult(np.nan, np.nan, np.nan, None)
 
     if contains_nan:
@@ -3920,10 +3966,11 @@ def median_test(*samples, ties='below', correction=True, lambda_=1,
         # check for that case here.
         zero_cols = np.nonzero((table == 0).all(axis=0))[0]
         if len(zero_cols) > 0:
-            msg = ("All values in sample %d are equal to the grand "
-                   "median (%r), so they are ignored, resulting in an "
-                   "empty sample." % (zero_cols[0] + 1, grand_median))
-            raise ValueError(msg)
+            raise ValueError(
+                f"All values in sample {zero_cols[0] + 1} are equal to the grand "
+                f"median ({grand_median!r}), so they are ignored, resulting in an "
+                f"empty sample."
+            )
 
     stat, p, dof, expected = chi2_contingency(table, lambda_=lambda_,
                                               correction=correction)
@@ -3933,9 +3980,7 @@ def median_test(*samples, ties='below', correction=True, lambda_=1,
 def _circfuncs_common(samples, period, xp=None):
     xp = array_namespace(samples) if xp is None else xp
 
-    if xp.isdtype(samples.dtype, 'integral'):
-        dtype = xp.asarray(1.).dtype  # get default float type
-        samples = xp.asarray(samples, dtype=dtype)
+    samples = xp_promote(samples, force_floating=True, xp=xp)
 
     # Recast samples as radians that range between 0 and 2 pi and calculate
     # the sine and cosine
@@ -3948,7 +3993,7 @@ def _circfuncs_common(samples, period, xp=None):
 
 @_axis_nan_policy_factory(
     lambda x: x, n_outputs=1, default_axis=None,
-    result_to_tuple=lambda x: (x,)
+    result_to_tuple=lambda x, _: (x,)
 )
 def circmean(samples, high=2*pi, low=0, axis=None, nan_policy='propagate'):
     r"""Compute the circular mean of a sample of angle observations.
@@ -4041,7 +4086,7 @@ def circmean(samples, high=2*pi, low=0, axis=None, nan_policy='propagate'):
 
 @_axis_nan_policy_factory(
     lambda x: x, n_outputs=1, default_axis=None,
-    result_to_tuple=lambda x: (x,)
+    result_to_tuple=lambda x, _: (x,)
 )
 def circvar(samples, high=2*pi, low=0, axis=None, nan_policy='propagate'):
     r"""Compute the circular variance of a sample of angle observations.
@@ -4135,7 +4180,7 @@ def circvar(samples, high=2*pi, low=0, axis=None, nan_policy='propagate'):
 
 @_axis_nan_policy_factory(
     lambda x: x, n_outputs=1, default_axis=None,
-    result_to_tuple=lambda x: (x,)
+    result_to_tuple=lambda x, _: (x,)
 )
 def circstd(samples, high=2*pi, low=0, axis=None, nan_policy='propagate', *,
             normalize=False):

scipy/stats/_mstats_basic.py

@@ -2263,8 +2263,8 @@ def trimmed_stde(a, limits=(0.1,0.1), inclusive=(1,1), axis=None):
         return _trimmed_stde_1D(a.ravel(),lolim,uplim,loinc,upinc)
     else:
         if a.ndim > 2:
-            raise ValueError("Array 'a' must be at most two dimensional, "
-                             "but got a.ndim = %d" % a.ndim)
+            raise ValueError(f"Array 'a' must be at most two dimensional, "
+                             f"but got a.ndim = {a.ndim}")
         return ma.apply_along_axis(_trimmed_stde_1D, axis, a,
                                    lolim,uplim,loinc,upinc)
 
@@ -2641,7 +2641,7 @@ def winsorize(a, limits=None, inclusive=(True, True), inplace=False,
                 a[idx[upidx:]] = a[idx[upidx - 1]]
         return a
 
-    contains_nan, nan_policy = _contains_nan(a, nan_policy)
+    contains_nan = _contains_nan(a, nan_policy)
     # We are going to modify a: better make a copy
     a = ma.array(a, copy=np.logical_not(inplace))
 
@@ -3016,8 +3016,8 @@ def stde_median(data, axis=None):
         return _stdemed_1D(data)
     else:
         if data.ndim > 2:
-            raise ValueError("Array 'data' must be at most two dimensional, "
-                             "but got data.ndim = %d" % data.ndim)
+            raise ValueError(f"Array 'data' must be at most two dimensional, "
+                             f"but got data.ndim = {data.ndim}")
         return ma.apply_along_axis(_stdemed_1D, axis, data)
 
 
@@ -3067,9 +3067,8 @@ def skewtest(a, axis=0, alternative='two-sided'):
     b2 = skew(a,axis)
     n = a.count(axis)
     if np.min(n) < 8:
-        raise ValueError(
-            "skewtest is not valid with less than 8 samples; %i samples"
-            " were given." % np.min(n))
+        raise ValueError(f"skewtest is not valid with less than 8 samples; "
+                         f"{np.min(n)} samples were given.")
 
     y = b2 * ma.sqrt(((n+1)*(n+3)) / (6.0*(n-2)))
     beta2 = (3.0*(n*n+27*n-70)*(n+1)*(n+3)) / ((n-2.0)*(n+5)*(n+7)*(n+9))
@@ -3125,14 +3124,11 @@ def kurtosistest(a, axis=0, alternative='two-sided'):
     a, axis = _chk_asarray(a, axis)
     n = a.count(axis=axis)
     if np.min(n) < 5:
-        raise ValueError(
-            "kurtosistest requires at least 5 observations; %i observations"
-            " were given." % np.min(n))
+        raise ValueError(f"kurtosistest requires at least 5 observations; "
+                         f"{np.min(n)} observations were given.")
     if np.min(n) < 20:
-        warnings.warn(
-            "kurtosistest only valid for n>=20 ... continuing anyway, n=%i" % np.min(n),
-            stacklevel=2,
-        )
+        warnings.warn(f"kurtosistest only valid for n>=20 ... continuing "
+                      f"anyway, n={np.min(n)}", stacklevel=2)
 
     b2 = kurtosis(a, axis, fisher=False)
     E = 3.0*(n-1) / (n+1)
@@ -3535,8 +3531,8 @@ def friedmanchisquare(*args):
     data = argstoarray(*args).astype(float)
     k = len(data)
     if k < 3:
-        raise ValueError("Less than 3 groups (%i): " % k +
-                         "the Friedman test is NOT appropriate.")
+        raise ValueError(f"Less than 3 groups ({k}): the Friedman test "
+                         f"is NOT appropriate.")
 
     ranked = ma.masked_values(rankdata(data, axis=0), 0)
     if ranked._mask is not nomask:

scipy/stats/_mstats_extras.py

@@ -114,8 +114,8 @@ def hdquantiles(data, prob=(.25, .5, .75), axis=None, var=False,):
         result = _hd_1D(data, p, var)
     else:
         if data.ndim > 2:
-            raise ValueError("Array 'data' must be at most two dimensional, "
-                             "but got data.ndim = %d" % data.ndim)
+            raise ValueError(f"Array 'data' must be at most two dimensional, "
+                             f"but got data.ndim = {data.ndim}")
         result = ma.apply_along_axis(_hd_1D, axis, data, p, var)
 
     return ma.fix_invalid(result, copy=False)
@@ -203,8 +203,8 @@ def hdquantiles_sd(data, prob=(.25, .5, .75), axis=None):
         result = _hdsd_1D(data, p)
     else:
         if data.ndim > 2:
-            raise ValueError("Array 'data' must be at most two dimensional, "
-                             "but got data.ndim = %d" % data.ndim)
+            raise ValueError(f"Array 'data' must be at most two dimensional, "
+                             f"but got data.ndim = {data.ndim}")
         result = ma.apply_along_axis(_hdsd_1D, axis, data, p)
 
     return ma.fix_invalid(result, copy=False).ravel()
@@ -295,8 +295,8 @@ def mjci(data, prob=(0.25, 0.5, 0.75), axis=None):
 
     data = ma.array(data, copy=False)
     if data.ndim > 2:
-        raise ValueError("Array 'data' must be at most two dimensional, "
-                         "but got data.ndim = %d" % data.ndim)
+        raise ValueError(f"Array 'data' must be at most two dimensional, "
+                         f"but got data.ndim = {data.ndim}")
 
     p = np.atleast_1d(np.asarray(prob))
     # Computes quantiles along axis (or globally)
@@ -384,8 +384,8 @@ def median_cihs(data, alpha=0.05, axis=None):
         result = _cihs_1D(data, alpha)
     else:
         if data.ndim > 2:
-            raise ValueError("Array 'data' must be at most two dimensional, "
-                             "but got data.ndim = %d" % data.ndim)
+            raise ValueError(f"Array 'data' must be at most two dimensional, "
+                             f"but got data.ndim = {data.ndim}")
         result = ma.apply_along_axis(_cihs_1D, axis, data, alpha)
 
     return result