Tutorial and Example bug fixes (#83)

* adressing bug in tutorial * addressing future warning * addressing test warnings
sandialabs · Oct 24, 2023 · 86ad883 · 86ad883
1 parent 28778d1
commit 86ad883
Show file tree

Hide file tree

Showing 5 changed files with 560 additions and 514 deletions.
diff --git a/pvops/text/defaults.py b/pvops/text/defaults.py
@@ -95,7 +95,6 @@ def supervised_classifier_defs(settings_flag):
             },
             "RidgeClassifier": {
                 "clf__alpha": [0.0, 1e-3, 1.0],
-                "clf__normalize": [False, True],
             },
             "SGDClassifier": {
                 "clf__loss": ["squared_hinge"],
@@ -158,7 +157,7 @@ def supervised_classifier_defs(settings_flag):
             },
             "RidgeClassifier": {
                 "clf__alpha": [0.0, 1e-3, 1.0],
-                "clf__normalize": [False, True],
+                # "clf__normalize": [False, True],
             },
             "SGDClassifier": {
                 "clf__loss": ["squared_hinge"],
@@ -221,7 +220,7 @@ def supervised_classifier_defs(settings_flag):
             },
             "RidgeClassifier": {
                 "clf__alpha": [0.0, 1e-3, 1.0, 1e-4, 1e-3, 1e-2, 1e-1, 1.0],
-                "clf__normalize": [False, True],
+                # "clf__normalize": [False, True],
             },
             "SGDClassifier": {
                 "clf__loss": ["squared_hinge", "hinge", "log"],

diff --git a/pvops/text/visualize.py b/pvops/text/visualize.py
@@ -178,7 +178,7 @@ def restructure(vals, inds, ind_set):
     newdf[LABEL_COLUMN] = asset_sums
     newdf[DATE_COLUMN] = index_sums
 
-    cmap = plt.cm.get_cmap(cmap_name, len(asset_set))
+    cmap = matplotlib.colormaps.get_cmap(cmap_name).resampled(len(asset_set))
 
     graphs = []
     for i, a in enumerate(asset_set):

diff --git a/pvops/text2time/utils.py b/pvops/text2time/utils.py
@@ -284,6 +284,8 @@ def om_summary_stats(om_df, meta_df, om_col_dict, meta_col_dict):
     # Adding age column to om_df, but first initiating a COD column in the
     # OM-data (using NANs) to be able to take the difference between two columns
     om_df[meta_cod] = np.nan
+    om_df[meta_cod] = om_df[meta_cod].astype("O")
+
     for i in cod_dates.index:
         om_df.loc[i, meta_cod] = cod_dates[i]
     om_df[meta_cod] = pd.to_datetime(om_df[meta_cod])
@@ -485,7 +487,7 @@ def prod_anomalies(prod_df, prod_col_dict, minval=1.0, repval=np.nan, ffill=True
     prod_df.loc[mask, prod_ener] = repval
 
     if ffill:
-        prod_df.loc[:, prod_ener].fillna(method="ffill", inplace=True)
+        prod_df.loc[:, prod_ener].ffill(inplace=True)
         addressed = addressedwna
 
     return prod_df, addressed

diff --git a/tutorials/text_class_example.py b/tutorials/text_class_example.py
@@ -50,7 +50,7 @@ def visualize_attribute_timeseries(self, DATE_COLUMN):
 
     def visualize_cluster_entropy(self, cols):
         def eval_kmeans(X, k):
-            km = KMeans(n_clusters=k)
+            km = KMeans(n_clusters=k, n_init=10)
             km.fit(X)
             return km
 

diff --git a/tutorials/tutorial_textmodule.ipynb b/tutorials/tutorial_textmodule.ipynb