formatting

Author: glemaitre

content/python_files/multiple_horizons_prediction.py

@@ -322,4 +322,3 @@ def scoring(regressor, X, y):
 # of scores or computational cost. The trend of the scores along the horizon is also
 # different from the `HistGradientBoostingRegressor`: the scores worsen as the horizon
 # increases.
-

content/python_files/prediction_intervals.py

@@ -247,9 +247,7 @@
 
 # %%
 plot_residuals_vs_predicted(cv_predictions_hgbr_50).interactive().properties(
-    title=(
-        "Residuals vs Predicted Values from cross-validation predictions for median"
-    )
+    title=("Residuals vs Predicted Values from cross-validation predictions for median")
 )
 
 # %%
@@ -315,6 +313,7 @@
 # Now, we assess if the actual coverage of the models is close to the target coverage of
 # 90%. In addition, we compute the average width of the bands.
 
+
 # %%
 def coverage(y_true, y_quantile_low, y_quantile_high):
     y_true = np.asarray(y_true)
@@ -384,7 +383,11 @@ def mean_width(y_true, y_quantile_low, y_quantile_high):
 # %%
 ax = coverage_by_bin.boxplot(column="coverage", by="bin_label", whis=1000)
 ax.axhline(y=0.9, color="red", linestyle="--", label="Target coverage (0.9)")
-ax.set(xlabel="Load bins (MW)", ylabel="Coverage", title="Coverage Distribution by Load Bins")
+ax.set(
+    xlabel="Load bins (MW)",
+    ylabel="Coverage",
+    title="Coverage Distribution by Load Bins",
+)
 ax.set_title("Coverage Distribution by Load Bins")
 ax.legend()
 plt.suptitle("")  # Remove automatic suptitle from boxplot

content/python_files/single_horizon_prediction.py

@@ -271,12 +271,12 @@
 # Those results show very good performance of the model: less than 3% of mean
 # absolute percentage error (MAPE) on the test folds. Similarly, all the
 # deviance scores are close to 1.0.
-# 
+#
 # We observe a bit of variability in the scores across the different folds: in
 # particular the test performance on the first fold seems to be worse than the
 # other folds. This is likely due to the fact that the first fold contains
 # training data from 2021 and 2022 and the test data mostly from 2023.
-# 
+#
 # The invasion in Ukraine and a sharp drop in nuclear electricity production
 # due to safety problems strongly impacted the distribution of the electricity
 # prices in 2022, with unprecedented high prices, which can in turn cause a
@@ -642,4 +642,3 @@
 
 # %%
 # nested_cv_results_ridge.round(3)
-