Add support for loading PLS model correctly on python 3.7 or sklearn …

…< 1.1

feat/plotting.py

@@ -991,7 +991,7 @@ def predict(au, model=None, feature_range=None):
     if len(au) != model.n_components:
         print(au)
         print(model.n_components)
-        raise ValueError("au vector must be len(", model.n_components, ").")
+        raise ValueError(f"au vector must be length {model.n_components}.")
 
     if len(au.shape) == 1:
         au = np.reshape(au, (1, -1))

feat/utils.py

@@ -262,10 +262,17 @@ def load_viz_model(
             hf = h5py.File(h5_path, mode="r")
             x_weights = np.array(hf.get("x_weights"))
             model = PLSRegression(n_components=x_weights.shape[1])
-            # PLSRegression in < 1.3 stores coefs ax features x samples unlike other
-            # estimators
-            model.__dict__["coef_"] = np.array(hf.get("coef")).T
-            model.__dict__["_coef_"] = np.array(hf.get("coef")).T
+            # PLSRegression in sklearn < 1.1 storex coefs as samples x features, but
+            # recent versions transpose this. Check if the user is on Python 3.7 (which
+            # only supports sklearn 1.0.x) or < sklearn 1.1.x
+            if (my_pymajor == 3 and my_pyminor == 7) or (
+                my_skmajor == 1 and my_skminor != 1
+            ):
+                model.__dict__["coef_"] = np.array(hf.get("coef"))
+                model.__dict__["_coef_"] = np.array(hf.get("coef"))
+            else:
+                model.__dict__["coef_"] = np.array(hf.get("coef")).T
+                model.__dict__["_coef_"] = np.array(hf.get("coef")).T
             model.__dict__["x_weights_"] = np.array(hf.get("x_weights"))
             model.__dict__["y_weights_"] = np.array(hf.get("y_weights"))
             model.__dict__["x_loadings"] = np.array(hf.get("x_loadings"))

requirements.txt

@@ -3,7 +3,7 @@ numpy>=1.9
 seaborn>=0.7.0
 matplotlib>=2.1
 nltools>=0.3.6
-scikit-learn>=1.1
+scikit-learn>=1.0
 pywavelets>=0.3.0
 h5py>=2.7.0
 Pillow>=6.0.0