Merge pull request #637 from bashtage/improve-typing-v2

bashtage · web-flow · commit 775ce62b63fc · 2025-03-22T18:43:26.000Z
TYP: Typing improvements
diff --git a/examples/asset-pricing_examples.ipynb b/examples/asset-pricing_examples.ipynb
@@ -412,7 +412,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/asset-pricing_formulas.ipynb b/examples/asset-pricing_formulas.ipynb
@@ -139,7 +139,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/iv_absorbing-regression.ipynb b/examples/iv_absorbing-regression.ipynb
@@ -151,7 +151,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/iv_advanced-examples.ipynb b/examples/iv_advanced-examples.ipynb
@@ -570,7 +570,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "nbsphinx": {
    "allow_errors": true
diff --git a/examples/iv_basic-examples.ipynb b/examples/iv_basic-examples.ipynb
@@ -805,7 +805,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/iv_using-formulas.ipynb b/examples/iv_using-formulas.ipynb
@@ -191,7 +191,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/panel_data-formats.ipynb b/examples/panel_data-formats.ipynb
@@ -230,7 +230,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/panel_examples.ipynb b/examples/panel_examples.ipynb
@@ -429,7 +429,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/panel_using-formulas.ipynb b/examples/panel_using-formulas.ipynb
@@ -182,7 +182,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/system_examples.ipynb b/examples/system_examples.ipynb
@@ -779,7 +779,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/system_formulas.ipynb b/examples/system_formulas.ipynb
@@ -180,7 +180,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/examples/system_three-stage-ls.ipynb b/examples/system_three-stage-ls.ipynb
@@ -353,7 +353,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.7"
+   "version": "3.12.9"
   },
   "pycharm": {
    "stem_cell": {
diff --git a/linearmodels/asset_pricing/covariance.py b/linearmodels/asset_pricing/covariance.py
@@ -192,7 +192,7 @@ def cov(self) -> linearmodels.typing.data.Float64Array:
             out = ji @ s @ ji.T
         else:
             j = self.jacobian
-            out = inv(j.T @ inv(s) @ j)
+            out = inv(j.T @ inv(s) @ j).astype(float, copy=False)
         out = (scale / 2) * (out + out.T)
         return out
 
@@ -371,4 +371,4 @@ def w(
             moments = moments - moments.mean(0)[None, :]
         out = self._kernel_cov(moments)
 
-        return inv(out)
+        return inv(out).astype(float, copy=False)
diff --git a/linearmodels/asset_pricing/model.py b/linearmodels/asset_pricing/model.py
@@ -442,7 +442,7 @@ def __init__(
             self._sigma = np.asarray(sigma)
             vals, vecs = np.linalg.eigh(sigma)
             self._sigma_m12 = vecs @ np.diag(1.0 / np.sqrt(vals)) @ vecs.T
-            self._sigma_inv = np.linalg.inv(self._sigma)
+            self._sigma_inv = np.linalg.inv(self._sigma).astype(float, copy=False)
 
     def __str__(self) -> str:
         out = super().__str__()
@@ -966,10 +966,10 @@ def fit(
             self.portfolios, self.factors, risk_free=self._risk_free
         )
         res = mod.fit()
-        betas = np.asarray(res.betas).ravel()
+        betas_1d = np.asarray(res.betas).ravel()
         lam = np.asarray(res.risk_premia)
         mu = self.factors.ndarray.mean(0)
-        sv = np.r_[betas, lam, mu][:, None]
+        sv = np.r_[betas_1d, lam, mu][:, None]
         if starting is not None:
             starting = np.asarray(starting)
             if starting.ndim == 1:
diff --git a/linearmodels/iv/__init__.py b/linearmodels/iv/__init__.py
@@ -1,6 +1,6 @@
-from .absorbing import AbsorbingLS, Interaction  # flake8: noqa
-from .model import IV2SLS, IVGMM, IVGMMCUE, IVLIML  # flake8: noqa
-from .results import compare  # flake8: noqa
+from .absorbing import AbsorbingLS, Interaction
+from .model import IV2SLS, IVGMM, IVGMMCUE, IVLIML
+from .results import compare
 
 __all__ = [
     "IV2SLS",
diff --git a/linearmodels/iv/absorbing.py b/linearmodels/iv/absorbing.py
@@ -913,10 +913,10 @@ def _first_time_fit(
         self._regressors_hash = areg.hash
         self._constant_absorbed = self._has_constant_exog and areg_constant
 
-        dep = self._dependent.ndarray
+        dep = self._dependent.ndarray.astype(float, copy=False)
         exog = cast(linearmodels.typing.data.Float64Array, self._exog.ndarray)
 
-        root_w = sqrt(self._weight_data.ndarray)
+        root_w = sqrt(self._weight_data.ndarray.astype(float, copy=False))
         dep = root_w * dep
         exog = root_w * exog
         denom = root_w.T @ root_w
@@ -931,13 +931,13 @@ def _first_time_fit(
 
                 absorb_options["drop_singletons"] = False
                 algo = create(self._absorb_inter.cat, **absorb_options)
-                dep_exog = column_stack((dep, exog))
+                dep_exog = column_stack((dep, exog)).astype(float, copy=False)
                 resids = algo.residualize(dep_exog)
                 dep_resid = resids[:, :1]
                 exog_resid = resids[:, 1:]
             else:
                 self._regressors = preconditioner(self._regressors)[0]
-                dep_exog = column_stack((dep, exog))
+                dep_exog = column_stack((dep, exog)).astype(float, copy=False)
                 resid = lsmr_annihilate(
                     self._regressors,
                     dep_exog,
diff --git a/linearmodels/iv/covariance.py b/linearmodels/iv/covariance.py
@@ -104,8 +104,8 @@ def kernel_weight_quadratic_spectral(
         w[0] = 0
         return w
 
-    z = arange(n + 1) / float(bw)
-    w = 6 * pi * z / 5
+    z = arange(n + 1).astype(float) / float(bw)
+    w = cast(linearmodels.typing.data.FloatArray1D, 6 * pi * z / 5)
     w[0] = 1
     w[1:] = 3 / w[1:] ** 2 * (sin(w[1:]) / w[1:] - cos(w[1:]))
 
diff --git a/linearmodels/iv/results.py b/linearmodels/iv/results.py
@@ -710,9 +710,9 @@ def diagnostics(self) -> DataFrame:
 
         endog, exog, instr, weights = self.endog, self.exog, self.instr, self.weights
         w = sqrt(weights.ndarray)
-        z = w * instr.ndarray
+        z = w * instr.ndarray.astype(float, copy=False)
         nz = z.shape[1]
-        x = w * exog.ndarray
+        x = w * exog.ndarray.astype(float, copy=False)
         ez = annihilate(z, x)
         individual_results = self.individual
         out_df = DataFrame(
diff --git a/linearmodels/panel/data.py b/linearmodels/panel/data.py
@@ -195,9 +195,15 @@ def __init__(
                     if x.ndim == 2:
                         x = x.to_pandas()
                     else:
-                        items: list[Hashable] = np.asarray(x.coords[x.dims[0]]).tolist()
-                        major: list[Hashable] = np.asarray(x.coords[x.dims[1]]).tolist()
-                        minor: list[Hashable] = np.asarray(x.coords[x.dims[2]]).tolist()
+                        items: list[Hashable] = cast(
+                            list[Hashable], np.asarray(x.coords[x.dims[0]]).tolist()
+                        )
+                        major: list[Hashable] = cast(
+                            list[Hashable], np.asarray(x.coords[x.dims[1]]).tolist()
+                        )
+                        minor: list[Hashable] = cast(
+                            list[Hashable], np.asarray(x.coords[x.dims[2]]).tolist()
+                        )
                         values = x.values
                         x = panel_to_frame(values, items, major, minor, True)
             except ImportError:
@@ -538,7 +544,7 @@ def demean(  # noqa: E704
         group: Literal["entity", "time", "both"],
         weights: PanelData | None,
         return_panel: Literal[False],
-    ) -> linearmodels.typing.data.Float64Array: ...  # noqa: E704
+    ) -> linearmodels.typing.data.Float64Array: ...
 
     def demean(
         self,
diff --git a/linearmodels/panel/model.py b/linearmodels/panel/model.py
@@ -54,6 +54,7 @@
 from linearmodels.shared.linalg import has_constant
 from linearmodels.shared.typed_getters import get_panel_data_like
 from linearmodels.shared.utility import AttrDict, ensure_unique_column, panel_to_frame
+from linearmodels.typing import BoolArray
 import linearmodels.typing.data
 
 CovarianceEstimator = Union[
@@ -319,7 +320,9 @@ def __init__(
         self._is_weighted = True
         self._name = self.__class__.__name__
         self.weights = self._adapt_weights(weights)
-        self._not_null = np.ones(self.dependent.values2d.shape[0], dtype=bool)
+        self._not_null: BoolArray = np.ones(
+            self.dependent.values2d.shape[0], dtype=bool
+        )
         self._cov_estimators = CovarianceManager(
             self.__class__.__name__,
             HomoskedasticCovariance,
@@ -466,10 +469,11 @@ def _validate_data(self) -> None:
             )
 
         all_missing = np.any(np.isnan(y), axis=1) & np.all(np.isnan(x), axis=1)
-        missing = (
+        missing = np.asarray(
             np.any(np.isnan(y), axis=1)
             | np.any(np.isnan(x), axis=1)
-            | np.any(np.isnan(w), axis=1)
+            | np.any(np.isnan(w), axis=1),
+            dtype=bool,
         )
 
         missing_warning(np.asarray(all_missing ^ missing), stacklevel=4)
@@ -479,7 +483,7 @@ def _validate_data(self) -> None:
             self.weights.drop(missing)
 
             x = cast(linearmodels.typing.data.Float64Array, self.exog.values2d)
-            self._not_null = np.asarray(~missing)
+            self._not_null = cast(BoolArray, np.asarray(~missing))
 
         w_df = self.weights.dataframe
         if np.any(np.asarray(w_df) <= 0):
@@ -778,7 +782,9 @@ def _setup_clusters(
         if cluster_entity:
             group_ids_arr = self.dependent.entity_ids.squeeze()
             name = "cov.cluster.entity"
-            group_ids = Series(group_ids_arr, index=self.dependent.index, name=name)
+            group_ids: Series[int] = Series(
+                group_ids_arr, index=self.dependent.index, name=name
+            )
             if clusters_frame is not None:
                 clusters_frame[name] = group_ids
             else:
@@ -2021,7 +2027,7 @@ def fit(
             linearmodels.typing.data.Float64Array, np.sqrt(self.weights.values2d)
         )
         y_ex = root_w * self.dependent.values2d
-        mu_ex = 0
+        mu_ex = np.array(0.0, dtype=float)
         if (
             self.has_constant
             or self.entity_effects
@@ -2469,7 +2475,9 @@ def _setup_clusters(
         if cluster_entity:
             group_ids = self.dependent.entity_ids.squeeze()
             name = "cov.cluster.entity"
-            group_ids_s = Series(group_ids, index=self.dependent.index, name=name)
+            group_ids_s: Series[int] = Series(
+                group_ids, index=self.dependent.index, name=name
+            )
             if clusters_frame is not None:
                 clusters_frame[name] = group_ids_s
             else:
diff --git a/linearmodels/system/_utility.py b/linearmodels/system/_utility.py
@@ -308,7 +308,11 @@ def _compute_transform(self) -> None:
         t, left = vecs[:, : k - c], vecs[:, k - c :]
         q = self._qa[:, None]
         a = q.T @ inv(left.T @ r.T) @ left.T
-        self._t, self._l, self._a = t, left, a
+        self._t, self._l, self._a = (
+            cast(linearmodels.typing.data.FloatArray2D, t),
+            cast(linearmodels.typing.data.FloatArray2D, left),
+            cast(linearmodels.typing.data.FloatArray2D, a),
+        )
         self._computed = True
 
     @property
diff --git a/linearmodels/system/covariance.py b/linearmodels/system/covariance.py
@@ -129,7 +129,7 @@ def _mvreg_cov(self) -> linearmodels.typing.data.Float64Array:
     def _gls_cov(self) -> linearmodels.typing.data.Float64Array:
         x = self._x
         sigma = self._sigma
-        sigma_inv = inv(sigma)
+        sigma_inv = cast(linearmodels.typing.data.Float64Array, inv(sigma))
 
         xpx = blocked_inner_prod(x, sigma_inv)
         # Handles case where sigma_inv is not inverse of full_sigma
@@ -229,7 +229,7 @@ def __init__(
         nobs = eps.shape[0]
 
         if gls:
-            weights = inv(sigma)
+            weights = cast(linearmodels.typing.data.Float64Array, inv(sigma))
             bigx = blocked_diag_product(x, weights)
             e = eps.T.ravel()[:, None]
             bigxe = bigx * e
@@ -258,7 +258,9 @@ def _cov(self, gls: bool) -> linearmodels.typing.data.Float64Array:
         nobs = x[0].shape[0]
         k = len(x)
         sigma = self.sigma
-        weights = inv(sigma) if gls else eye(k)
+        weights = (
+            cast(linearmodels.typing.data.Float64Array, inv(sigma)) if gls else eye(k)
+        )
         xpx = blocked_inner_prod(x, weights) / nobs
         xeex = self._xeex()
 
diff --git a/linearmodels/system/model.py b/linearmodels/system/model.py
@@ -146,7 +146,7 @@ def _parameters_from_xprod(
         params_c = solve(xpx, xpy)
         params = cons.t @ params_c + cons.a.T
     else:
-        params = solve(xpx, xpy)
+        params = solve(xpx, xpy).astype(float, copy=False)
     return params
 
 
@@ -760,7 +760,7 @@ def _gls_estimate(
 
         if not full_cov:
             sigma = np.diag(np.diag(sigma))
-        sigma_inv = inv(sigma)
+        sigma_inv = cast(linearmodels.typing.data.Float64Array, inv(sigma))
 
         k = len(wy)
 
@@ -1045,7 +1045,7 @@ def _system_r2(
             est_sigma = sigma
             if not full_cov:
                 est_sigma = np.diag(np.diag(est_sigma))
-        est_sigma_inv = inv(est_sigma)
+        est_sigma_inv = cast(linearmodels.typing.data.Float64Array, inv(est_sigma))
         nobs = wy[0].shape[0]
         k = len(wy)
         xpx = blocked_inner_prod(wi, est_sigma_inv)
diff --git a/linearmodels/typing/__init__.py b/linearmodels/typing/__init__.py
diff --git a/linearmodels/typing/data.py b/linearmodels/typing/data.py
diff --git a/requirements-dev.txt b/requirements-dev.txt
