Small fixes found we stopped using pre-expansion weights

.github/workflows/python-app.yml

@@ -21,7 +21,7 @@ jobs:
     - uses: actions/checkout@v6
 
     - name: Install uv
-      uses: astral-sh/setup-uv@v8.0.0
+      uses: astral-sh/setup-uv@v8.1.0
 
     - name: Set up Python ${{ matrix.python-version }}
       run: uv python install ${{ matrix.python-version }}

pySEQTarget/error/_data_checker.py

@@ -18,6 +18,10 @@ def _data_checker(self):
 
     for col in self.weight_eligible_colnames:
         if col is not None:
+            if col not in self.data.columns:
+                raise ValueError(
+                    f"weight_eligible_colnames entry '{col}' not found in data columns."
+                )
             _check_binary(self.data, col)
 
     check = self.data.group_by(self.id_col).agg(

pySEQTarget/expansion/_binder.py

@@ -89,10 +89,14 @@ def _binder(self, kept_cols):
         for c in baseline_cols
     ]
 
+    to_drop = [f"{self.eligible_col}{self.indicator_baseline}"]
+    if self.eligible_col not in kept_cols:
+        to_drop.append(self.eligible_col)
+
     DT = (
         DT.with_columns(bas)
         .filter(pl.col(f"{self.eligible_col}{self.indicator_baseline}") == 1)
-        .drop([f"{self.eligible_col}{self.indicator_baseline}", self.eligible_col])
+        .drop(to_drop)
     )
 
     # Truncate each (id, trial) at the first outcome event so that subjects who

pySEQTarget/helpers/_bootstrap.py

@@ -1,5 +1,6 @@
 import copy
 import time
+import warnings
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from functools import wraps
 
@@ -103,7 +104,7 @@ def wrapper(self, *args, **kwargs):
                 self.DT = None
 
                 with ProcessPoolExecutor(max_workers=ncores) as executor:
-                    futures = [
+                    futures = {
                         executor.submit(
                             _bootstrap_worker,
                             self,
@@ -113,13 +114,24 @@ def wrapper(self, *args, **kwargs):
                             seed,
                             args,
                             kwargs,
-                        )
+                        ): i
                         for i in range(nboot)
-                    ]
+                    }
+                    skipped = 0
                     for j in tqdm(
                         as_completed(futures), total=nboot, desc="Bootstrapping..."
                     ):
-                        results.append(j.result())
+                        boot_idx = futures[j]
+                        try:
+                            results.append(j.result())
+                        except np.linalg.LinAlgError as e:
+                            skipped += 1
+                            warnings.warn(
+                                f"Bootstrap iteration {boot_idx + 1} failed "
+                                f"({e}); skipping replicate.",
+                                UserWarning,
+                                stacklevel=2,
+                            )
 
                 self._rng = original_rng
                 self.DT = self._offloader.load_dataframe(original_DT_ref)
@@ -131,6 +143,7 @@ def wrapper(self, *args, **kwargs):
                 else:
                     original_DT_ref = original_DT
 
+                skipped = 0
                 for i in tqdm(range(nboot), desc="Bootstrapping..."):
                     self._current_boot_idx = i + 1
                     if seed is not None:
@@ -140,12 +153,30 @@ def wrapper(self, *args, **kwargs):
                     if self._offloader.enabled:
                         del tmp
                     self.bootstrap_nboot = 0
-                    boot_fit = method(self, *args, **kwargs)
-                    results.append(boot_fit)
+                    try:
+                        boot_fit = method(self, *args, **kwargs)
+                        results.append(boot_fit)
+                    except np.linalg.LinAlgError as e:
+                        skipped += 1
+                        warnings.warn(
+                            f"Bootstrap iteration {i + 1} failed "
+                            f"({e}); skipping replicate.",
+                            UserWarning,
+                            stacklevel=2,
+                        )
 
-                self.bootstrap_nboot = nboot
                 self.DT = self._offloader.load_dataframe(original_DT_ref)
 
+            self.bootstrap_nboot = len(results) - 1
+            if skipped > 0:
+                warnings.warn(
+                    f"{skipped} of {nboot} bootstrap replicate(s) skipped due to "
+                    "singular Hessian; effective bootstrap_nboot is "
+                    f"{self.bootstrap_nboot}.",
+                    UserWarning,
+                    stacklevel=2,
+                )
+
             end = time.perf_counter()
             self._model_time = _format_time(start, end)
 

pySEQTarget/helpers/_predict_model.py

@@ -1,5 +1,3 @@
-import warnings
-
 import numpy as np
 
 from ._fix_categories import _fix_categories_for_predict

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "pySEQTarget"
-version = "0.12.9"
+version = "0.13.0"
 description = "Sequentially Nested Target Trial Emulation"
 readme = "README.md"
 license = {text = "MIT"}
@@ -83,3 +83,6 @@ SEQdata = ["data/*.csv"]
 [tool.pytest.ini_options]
 pythonpath = ["."]
 testpaths = ["tests"]
+filterwarnings = [
+    "ignore:FigureCanvasAgg is non-interactive:UserWarning",
+]

tests/test_plot.py

@@ -7,8 +7,8 @@
 
 matplotlib.use("Agg")  # non-interactive backend — no windows opened
 
-from pySEQTarget import SEQopts, SEQuential
-from pySEQTarget.data import load_data
+from pySEQTarget import SEQopts, SEQuential  # noqa: E402
+from pySEQTarget.data import load_data  # noqa: E402
 
 
 @pytest.fixture(autouse=True)