bartz 0.4.1__py3-none-any.whl → 0.6.0__py3-none-any.whl

bartz/BART.py

@@ -1,6 +1,6 @@
 # bartz/src/bartz/BART.py
 #
-# Copyright (c) 2024, Giacomo Petrillo
+# Copyright (c) 2024-2025, Giacomo Petrillo
 #
 # This file is part of bartz.
 #
@@ -22,16 +22,21 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
+"""Implement a user interface that mimics the R BART package."""
+
 import functools
+import math
+from typing import Any, Literal
 
 import jax
 import jax.numpy as jnp
+from jax.scipy.special import ndtri
+from jaxtyping import Array, Bool, Float, Float32
+
+from . import grove, jaxext, mcmcloop, mcmcstep, prepcovars
+
+FloatLike = float | Float[Any, '']
 
-from . import jaxext
-from . import grove
-from . import mcmcstep
-from . import mcmcloop
-from . import prepcovars
 
 class gbart:
     """
@@ -49,14 +54,18 @@ class gbart:
         The training responses.
     x_test : array (p, m) or DataFrame, optional
         The test predictors.
+    type
+        The type of regression. 'wbart' for continuous regression, 'pbart' for
+        binary regression with probit link.
     usequants : bool, default False
         Whether to use predictors quantiles instead of a uniform grid to bin
         predictors.
     sigest : float, optional
-        An estimate of the residual standard deviation on `y_train`, used to
-        set `lamda`. If not specified, it is estimated by linear regression.
-        If `y_train` has less than two elements, it is set to 1. If n <= p, it
-        is set to the variance of `y_train`. Ignored if `lamda` is specified.
+        An estimate of the residual standard deviation on `y_train`, used to set
+        `lamda`. If not specified, it is estimated by linear regression (with
+        intercept, and without taking into account `w`). If `y_train` has less
+        than two elements, it is set to 1. If n <= p, it is set to the standard
+        deviation of `y_train`. Ignored if `lamda` is specified.
     sigdf : int, default 3
         The degrees of freedom of the scaled inverse-chisquared prior on the
         noise variance.
@@ -72,16 +81,26 @@ class gbart:
         Parameters of the prior on tree node generation. The probability that a
         node at depth `d` (0-based) is non-terminal is ``base / (1 + d) **
         power``.
-    maxdepth : int, default 6
-        The maximum depth of the trees. This is 1-based, so with the default
-        ``maxdepth=6``, the depths of the levels range from 0 to 5.
-    lamda : float, optional
-        The scale of the prior on the noise variance. If ``lamda==1``, the
-        prior is an inverse chi-squared scaled to have harmonic mean 1. If
-        not specified, it is set based on `sigest` and `sigquant`.
-    offset : float, optional
+    lamda
+        The prior harmonic mean of the error variance. (The harmonic mean of x
+        is 1/mean(1/x).) If not specified, it is set based on `sigest` and
+        `sigquant`.
+    tau_num
+        The numerator in the expression that determines the prior standard
+        deviation of leaves. If not specified, default to ``(max(y_train) -
+        min(y_train)) / 2`` (or 1 if `y_train` has less than two elements) for
+        continuous regression, and 3 for binary regression.
+    offset
         The prior mean of the latent mean function. If not specified, it is set
-        to the mean of `y_train`. If `y_train` is empty, it is set to 0.
+        to the mean of `y_train` for continuous regression, and to
+        ``Phi^-1(mean(y_train))`` for binary regression. If `y_train` is empty,
+        `offset` is set to 0.
+    w : array (n,), optional
+        Coefficients that rescale the error standard deviation on each
+        datapoint. Not specifying `w` is equivalent to setting it to 1 for all
+        datapoints. Note: `w` is ignored in the automatic determination of
+        `sigest`, so either the weights should be O(1), or `sigest` should be
+        specified by the user.
     ntree : int, default 200
         The number of trees used to represent the latent mean function.
     numcut : int, default 255
@@ -104,10 +123,24 @@ class gbart:
         The number of initial MCMC samples to discard as burn-in.
     keepevery : int, default 1
         The thinning factor for the MCMC samples, after burn-in.
-    printevery : int, default 100
-        The number of iterations (including skipped ones) between each log.
+    printevery : int or None, default 100
+        The number of iterations (including thinned-away ones) between each log
+        line. Set to `None` to disable logging.
+
+        `printevery` has a few unexpected side effects. On cpu, interrupting
+        with ^C halts the MCMC only on the next log. And the total number of
+        iterations is a multiple of `printevery`, so if ``nskip + keepevery *
+        ndpost`` is not a multiple of `printevery`, some of the last iterations
+        will not be saved.
     seed : int or jax random key, default 0
         The seed for the random number generator.
+    maxdepth : int, default 6
+        The maximum depth of the trees. This is 1-based, so with the default
+        ``maxdepth=6``, the depths of the levels range from 0 to 5.
+    init_kw : dict
+        Additional arguments passed to `mcmcstep.init`.
+    run_mcmc_kw : dict
+        Additional arguments passed to `mcmcloop.run_mcmc`.
 
     Attributes
     ----------
@@ -125,22 +158,8 @@ class gbart:
         The standard deviation of the error in the burn-in phase.
     offset : float
         The prior mean of the latent mean function.
-    scale : float
-        The prior standard deviation of the latent mean function.
-    lamda : float
-        The prior harmonic mean of the error variance.
     sigest : float or None
         The estimated standard deviation of the error used to set `lamda`.
-    ntree : int
-        The number of trees.
-    maxdepth : int
-        The maximum depth of the trees.
-    initkw : dict
-        Additional arguments passed to `mcmcstep.init`.
-
-    Methods
-    -------
-    predict
 
     Notes
     -----
@@ -149,20 +168,27 @@ class gbart:
 
     - If `x_train` and `x_test` are matrices, they have one predictor per row
       instead of per column.
+    - If `type` is not specified, it is determined solely based on the data type
+      of `y_train`, and not on whether it contains only two unique values.
     - If ``usequants=False``, R BART switches to quantiles anyway if there are
       less predictor values than the required number of bins, while bartz
       always follows the specification.
     - The error variance parameter is called `lamda` instead of `lambda`.
     - `rm_const` is always `False`.
     - The default `numcut` is 255 instead of 100.
-    - A lot of functionality is missing (variable selection, discrete response).
+    - A lot of functionality is missing (e.g., variable selection).
     - There are some additional attributes, and some missing.
+    - The trees have a maximum depth.
 
-    The linear regression used to set `sigest` adds an intercept.
     """
 
-    def __init__(self, x_train, y_train, *,
+    def __init__(
+        self,
+        x_train,
+        y_train,
+        *,
         x_test=None,
+        type: Literal['wbart', 'pbart'] = 'wbart',
         usequants=False,
         sigest=None,
         sigdf=3,
@@ -170,9 +196,10 @@ class gbart:
         k=2,
         power=2,
         base=0.95,
-        maxdepth=6,
-        lamda=None,
-        offset=None,
+        lamda: FloatLike | None = None,
+        tau_num: FloatLike | None = None,
+        offset: FloatLike | None = None,
+        w=None,
         ntree=200,
         numcut=255,
         ndpost=1000,
@@ -180,36 +207,52 @@ class gbart:
         keepevery=1,
         printevery=100,
         seed=0,
-        initkw={},
-        ):
-
+        maxdepth=6,
+        init_kw=None,
+        run_mcmc_kw=None,
+    ):
         x_train, x_train_fmt = self._process_predictor_input(x_train)
-
-        y_train, y_train_fmt = self._process_response_input(y_train)
+        y_train, _ = self._process_response_input(y_train)
         self._check_same_length(x_train, y_train)
+        if w is not None:
+            w, _ = self._process_response_input(w)
+            self._check_same_length(x_train, w)
 
+        y_train = self._process_type_settings(y_train, type, w)
+        # from here onwards, the type is determined by y_train.dtype == bool
         offset = self._process_offset_settings(y_train, offset)
-        scale = self._process_scale_settings(y_train, k)
-        lamda, sigest = self._process_noise_variance_settings(x_train, y_train, sigest, sigdf, sigquant, lamda, offset)
+        sigma_mu = self._process_leaf_sdev_settings(y_train, k, ntree, tau_num)
+        lamda, sigest = self._process_error_variance_settings(
+            x_train, y_train, sigest, sigdf, sigquant, lamda
+        )
 
         splits, max_split = self._determine_splits(x_train, usequants, numcut)
         x_train = self._bin_predictors(x_train, splits)
 
-        y_train = self._transform_input(y_train, offset, scale)
-        lamda_scaled = lamda / (scale * scale)
-
-        mcmc_state = self._setup_mcmc(x_train, y_train, max_split, lamda_scaled, sigdf, power, base, maxdepth, ntree, initkw)
-        final_state, burnin_trace, main_trace = self._run_mcmc(mcmc_state, ndpost, nskip, keepevery, printevery, seed)
+        mcmc_state = self._setup_mcmc(
+            x_train,
+            y_train,
+            offset,
+            w,
+            max_split,
+            lamda,
+            sigma_mu,
+            sigdf,
+            power,
+            base,
+            maxdepth,
+            ntree,
+            init_kw,
+        )
+        final_state, burnin_trace, main_trace = self._run_mcmc(
+            mcmc_state, ndpost, nskip, keepevery, printevery, seed, run_mcmc_kw
+        )
 
-        sigma = self._extract_sigma(main_trace, scale)
-        first_sigma = self._extract_sigma(burnin_trace, scale)
+        sigma = self._extract_sigma(main_trace)
+        first_sigma = self._extract_sigma(burnin_trace)
 
-        self.offset = offset
-        self.scale = scale
-        self.lamda = lamda
+        self.offset = final_state.offset  # from the state because of buffer donation
         self.sigest = sigest
-        self.ntree = ntree
-        self.maxdepth = maxdepth
         self.sigma = sigma
         self.first_sigma = first_sigma
 
@@ -225,9 +268,8 @@ class gbart:
 
     @functools.cached_property
     def yhat_train(self):
-        x_train = self._mcmc_state['X']
-        yhat_train = self._predict(self._main_trace, x_train)
-        return self._transform_output(yhat_train, self.offset, self.scale)
+        x_train = self._mcmc_state.X
+        return self._predict(self._main_trace, x_train)
 
     @functools.cached_property
     def yhat_train_mean(self):
@@ -239,19 +281,26 @@ class gbart:
 
         Parameters
         ----------
-        x_test : array (m, p) or DataFrame
+        x_test : array (p, m) or DataFrame
             The test predictors.
 
         Returns
         -------
         yhat_test : array (ndpost, m)
             The conditional posterior mean at `x_test` for each MCMC iteration.
+
+        Raises
+        ------
+        ValueError
+            If `x_test` has a different format than `x_train`.
         """
         x_test, x_test_fmt = self._process_predictor_input(x_test)
-        self._check_compatible_formats(x_test_fmt, self._x_train_fmt)
+        if x_test_fmt != self._x_train_fmt:
+            raise ValueError(
+                f'Input format mismatch: {x_test_fmt=} != x_train_fmt={self._x_train_fmt!r}'
+            )
         x_test = self._bin_predictors(x_test, self._splits)
-        yhat_test = self._predict(self._main_trace, x_test)
-        return self._transform_output(yhat_test, self.offset, self.scale)
+        return self._predict(self._main_trace, x_test)
 
     @staticmethod
     def _process_predictor_input(x):
@@ -264,10 +313,6 @@ class gbart:
         assert x.ndim == 2
         return x, fmt
 
-    @staticmethod
-    def _check_compatible_formats(fmt1, fmt2):
-        assert fmt1 == fmt2
-
     @staticmethod
     def _process_response_input(y):
         if hasattr(y, 'to_numpy'):
@@ -285,20 +330,30 @@ class gbart:
         assert get_length(x1) == get_length(x2)
 
     @staticmethod
-    def _process_noise_variance_settings(x_train, y_train, sigest, sigdf, sigquant, lamda, offset):
-        if lamda is not None:
+    def _process_error_variance_settings(
+        x_train, y_train, sigest, sigdf, sigquant, lamda
+    ) -> tuple[Float32[Array, ''] | None, ...]:
+        if y_train.dtype == bool:
+            if sigest is not None:
+                raise ValueError('Let `sigest=None` for binary regression')
+            if lamda is not None:
+                raise ValueError('Let `lamda=None` for binary regression')
+            return None, None
+        elif lamda is not None:
+            if sigest is not None:
+                raise ValueError('Let `sigest=None` if `lamda` is specified')
             return lamda, None
         else:
             if sigest is not None:
-                sigest2 = sigest * sigest
+                sigest2 = jnp.square(sigest)
             elif y_train.size < 2:
                 sigest2 = 1
             elif y_train.size <= x_train.shape[0]:
-                sigest2 = jnp.var(y_train - offset)
+                sigest2 = jnp.var(y_train)
             else:
                 x_centered = x_train.T - x_train.mean(axis=1)
                 y_centered = y_train - y_train.mean()
-                    # centering is equivalent to adding an intercept column
+                # centering is equivalent to adding an intercept column
                 _, chisq, rank, _ = jnp.linalg.lstsq(x_centered, y_centered)
                 chisq = chisq.squeeze(0)
                 dof = len(y_train) - rank
@@ -309,20 +364,62 @@ class gbart:
             return sigest2 / invchi2rid, jnp.sqrt(sigest2)
 
     @staticmethod
-    def _process_offset_settings(y_train, offset):
+    def _process_type_settings(y_train, type, w):
+        match type:
+            case 'wbart':
+                if y_train.dtype != jnp.float32:
+                    raise TypeError(
+                        'Continuous regression requires y_train.dtype=float32,'
+                        f' got {y_train.dtype=} instead.'
+                    )
+            case 'pbart':
+                if w is not None:
+                    raise ValueError(
+                        'Binary regression does not support weights, set `w=None`'
+                    )
+                if y_train.dtype != bool:
+                    raise TypeError(
+                        'Binary regression requires y_train.dtype=bool,'
+                        f' got {y_train.dtype=} instead.'
+                    )
+            case _:
+                raise ValueError(f'Invalid {type=}')
+
+        return y_train
+
+    @staticmethod
+    def _process_offset_settings(
+        y_train: Float32[Array, 'n'] | Bool[Array, 'n'],
+        offset: float | Float32[Any, ''] | None,
+    ) -> Float32[Array, '']:
         if offset is not None:
-            return offset
+            return jnp.asarray(offset)
         elif y_train.size < 1:
-            return 0
+            return jnp.array(0.0)
         else:
-            return y_train.mean()
+            mean = y_train.mean()
 
-    @staticmethod
-    def _process_scale_settings(y_train, k):
-        if y_train.size < 2:
-            return 1
+        if y_train.dtype == bool:
+            return ndtri(mean)
         else:
-            return (y_train.max() - y_train.min()) / (2 * k)
+            return mean
+
+    @staticmethod
+    def _process_leaf_sdev_settings(
+        y_train: Float32[Array, 'n'] | Bool[Array, 'n'],
+        k: float,
+        ntree: int,
+        tau_num: FloatLike | None,
+    ):
+        if tau_num is None:
+            if y_train.dtype == bool:
+                tau_num = 3.0
+            elif y_train.size < 2:
+                tau_num = 1.0
+            else:
+                tau_num = (y_train.max() - y_train.min()) / 2
+
+        return tau_num / (k * math.sqrt(ntree))
 
     @staticmethod
     def _determine_splits(x_train, usequants, numcut):
@@ -336,52 +433,86 @@ class gbart:
         return prepcovars.bin_predictors(x, splits)
 
     @staticmethod
-    def _transform_input(y, offset, scale):
-        return (y - offset) / scale
-
-    @staticmethod
-    def _setup_mcmc(x_train, y_train, max_split, lamda, sigdf, power, base, maxdepth, ntree, initkw):
+    def _setup_mcmc(
+        x_train,
+        y_train,
+        offset,
+        w,
+        max_split,
+        lamda,
+        sigma_mu,
+        sigdf,
+        power,
+        base,
+        maxdepth,
+        ntree,
+        init_kw,
+    ):
         depth = jnp.arange(maxdepth - 1)
         p_nonterminal = base / (1 + depth).astype(float) ** power
-        sigma2_alpha = sigdf / 2
-        sigma2_beta = lamda * sigma2_alpha
+
+        if y_train.dtype == bool:
+            sigma2_alpha = None
+            sigma2_beta = None
+        else:
+            sigma2_alpha = sigdf / 2
+            sigma2_beta = lamda * sigma2_alpha
+
         kw = dict(
             X=x_train,
-            y=y_train,
+            # copy y_train because it's going to be donated in the mcmc loop
+            y=jnp.array(y_train),
+            offset=offset,
+            error_scale=w,
             max_split=max_split,
             num_trees=ntree,
             p_nonterminal=p_nonterminal,
+            sigma_mu2=jnp.square(sigma_mu),
             sigma2_alpha=sigma2_alpha,
             sigma2_beta=sigma2_beta,
             min_points_per_leaf=5,
         )
-        kw.update(initkw)
+        if init_kw is not None:
+            kw.update(init_kw)
         return mcmcstep.init(**kw)
 
     @staticmethod
-    def _run_mcmc(mcmc_state, ndpost, nskip, keepevery, printevery, seed):
-        if isinstance(seed, jax.Array) and jnp.issubdtype(seed.dtype, jax.dtypes.prng_key):
-            key = seed
+    def _run_mcmc(mcmc_state, ndpost, nskip, keepevery, printevery, seed, run_mcmc_kw):
+        if isinstance(seed, jax.Array) and jnp.issubdtype(
+            seed.dtype, jax.dtypes.prng_key
+        ):
+            key = seed.copy()
+            # copy because the inner loop in run_mcmc will donate the buffer
         else:
             key = jax.random.key(seed)
-        callback = mcmcloop.make_simple_print_callback(printevery)
-        return mcmcloop.run_mcmc(mcmc_state, nskip, ndpost, keepevery, callback, key)
 
-    @staticmethod
-    def _predict(trace, x):
-        return mcmcloop.evaluate_trace(trace, x)
+        kw = dict(
+            n_burn=nskip,
+            n_skip=keepevery,
+            inner_loop_length=printevery,
+            allow_overflow=True,
+        )
+        if printevery is not None:
+            kw.update(mcmcloop.make_print_callbacks())
+        if run_mcmc_kw is not None:
+            kw.update(run_mcmc_kw)
 
-    @staticmethod
-    def _transform_output(y, offset, scale):
-        return offset + scale * y
+        return mcmcloop.run_mcmc(key, mcmc_state, ndpost, **kw)
 
     @staticmethod
-    def _extract_sigma(trace, scale):
-        return scale * jnp.sqrt(trace['sigma2'])
+    def _extract_sigma(trace) -> Float32[Array, 'trace_length'] | None:
+        if trace['sigma2'] is None:
+            return None
+        else:
+            return jnp.sqrt(trace['sigma2'])
 
+    @staticmethod
+    def _predict(trace, x):
+        return mcmcloop.evaluate_trace(trace, x)
 
     def _show_tree(self, i_sample, i_tree, print_all=False):
         from . import debug
+
         trace = self._main_trace
         leaf_tree = trace['leaf_trees'][i_sample, i_tree]
         var_tree = trace['var_trees'][i_sample, i_tree]
@@ -396,30 +527,49 @@ class gbart:
         else:
             resid = bart['resid']
             alpha = bart['sigma2_alpha'] + resid.size / 2
-            norm2 = jnp.dot(resid, resid, preferred_element_type=bart['sigma2_beta'].dtype)
+            norm2 = jnp.dot(
+                resid, resid, preferred_element_type=bart['sigma2_beta'].dtype
+            )
             beta = bart['sigma2_beta'] + norm2 / 2
         sigma2 = beta / alpha
-        return jnp.sqrt(sigma2) * self.scale
+        return jnp.sqrt(sigma2)
 
     def _compare_resid(self):
         bart = self._mcmc_state
-        resid1 = bart['resid']
-        yhat = grove.evaluate_forest(bart['X'], bart['leaf_trees'], bart['var_trees'], bart['split_trees'], jnp.float32)
-        resid2 = bart['y'] - yhat
+        resid1 = bart.resid
+
+        trees = grove.evaluate_forest(
+            bart.X,
+            bart.forest.leaf_trees,
+            bart.forest.var_trees,
+            bart.forest.split_trees,
+            jnp.float32,  # TODO remove these configurable dtypes around
+        )
+
+        if bart.z is not None:
+            ref = bart.z
+        else:
+            ref = bart.y
+        resid2 = ref - (trees + bart.offset)
+
         return resid1, resid2
 
     def _avg_acc(self):
         trace = self._main_trace
+
         def acc(prefix):
             acc = trace[f'{prefix}_acc_count']
             prop = trace[f'{prefix}_prop_count']
             return acc.sum() / prop.sum()
+
         return acc('grow'), acc('prune')
 
     def _avg_prop(self):
         trace = self._main_trace
+
         def prop(prefix):
             return trace[f'{prefix}_prop_count'].sum()
+
         pgrow = prop('grow')
         pprune = prop('prune')
         total = pgrow + pprune
@@ -432,16 +582,19 @@ class gbart:
 
     def _depth_distr(self):
         from . import debug
+
         trace = self._main_trace
         split_trees = trace['split_trees']
         return debug.trace_depth_distr(split_trees)
 
     def _points_per_leaf_distr(self):
         from . import debug
-        return debug.trace_points_per_leaf_distr(self._main_trace, self._mcmc_state['X'])
+
+        return debug.trace_points_per_leaf_distr(self._main_trace, self._mcmc_state.X)
 
     def _check_trees(self):
         from . import debug
+
         return debug.check_trace(self._main_trace, self._mcmc_state)
 
     def _tree_goes_bad(self):

bartz/__init__.py

@@ -1,6 +1,6 @@
 # bartz/src/bartz/__init__.py
 #
-# Copyright (c) 2024, Giacomo Petrillo
+# Copyright (c) 2024-2025, Giacomo Petrillo
 #
 # This file is part of bartz.
 #
@@ -23,18 +23,10 @@
 # SOFTWARE.
 
 """
-Super-fast BART (Bayesian Additive Regression Trees) in Python
+Super-fast BART (Bayesian Additive Regression Trees) in Python.
 
 See the manual at https://gattocrucco.github.io/bartz/docs
 """
 
-from ._version import __version__
-
-from . import BART
-
-from . import debug
-from . import grove
-from . import mcmcstep
-from . import mcmcloop
-from . import prepcovars
-from . import jaxext
+from . import BART, debug, grove, jaxext, mcmcloop, mcmcstep, prepcovars  # noqa: F401
+from ._version import __version__  # noqa: F401

bartz/_version.py

@@ -1 +1 @@
-__version__ = '0.4.1'
+__version__ = '0.6.0'