Updates for review process plus README

Author: mariya

README.md

@@ -12,6 +12,8 @@
 
 <h1 class="custom-font miami-neon-text">Conformist</h1>
 
+## About
+
 Conformist v1.1.1 is an implementation of conformal prediction, specifically conformal risk control. It was written using Python 3.8.
 
 *BaseCoP* contains utility functions common to all conformal predictors, such as splitting data into calibration and validation sets, and setting up runs. It is extended by *FNRCoP* that implements conformal risk control.
@@ -68,3 +70,8 @@ mcp.calibrate()
 formatted_predictions = mcp.predict(testpd,
                                     OUTPUT_DIR)
 ```
+
+## Publication
+
+Mariya Lysenkova Wiklander et al. Error Reduction in Leukemia Machine Learning Classification With Conformal Prediction. JCO Clin Cancer Inform 9, e2400324(2025).
+DOI:[10.1200/CCI-24-00324](https://doi.org/10.1200/CCI-24-00324)

src/conformist/base_cop.py

@@ -20,6 +20,11 @@ class BaseCoP(OutputDir):
     FIGURE_HEIGHT = 8
     plt.rcParams.update({'font.size': FIGURE_FONTSIZE})
 
+    plt.rcParams["pdf.fonttype"] = 42  # If saving as PDF too
+    plt.rcParams["font.family"] = "Univers, DejaVu Sans"  # Choose a font that supports embedding
+
+
+
     def __init__(self, prediction_dataset: PredictionDataset, alpha=0.1):
         self.prediction_dataset = prediction_dataset
         self.alpha = alpha
@@ -198,8 +203,9 @@ def prediction_sets_to_text(self, prediction_sets, display_classes=None):
                 for prediction_set in prediction_sets]
 
     def upset_plot(self, predictions_sets, output_dir, color="black"):
-        plt.figure(figsize=(self.FIGURE_WIDTH,
-                            self.FIGURE_HEIGHT))
+        fig = plt.figure(figsize=(8, 3))
+        plt.tight_layout()
+        plt.rcParams.update({'font.size': 6})
 
         class_names = self.class_names
 
@@ -214,8 +220,17 @@ def upset_plot(self, predictions_sets, output_dir, color="black"):
         # Set a multi-index
         upset_data.set_index(upset_data.columns.tolist(), inplace=True)
 
-        plot(upset_data, sort_by="cardinality", facecolor=color, show_counts="%d", show_percentages="{:.0%}")
+        plot(upset_data,
+             fig=fig,
+             element_size=15,
+             sort_by="cardinality",
+             facecolor=color,
+             show_counts="%d",
+             show_percentages="{:.0%}")
+
+        # fig.set_size_inches(8, 3)
+        plt.tight_layout(w_pad=0)
 
-        path = f'{output_dir}/upset_plot.png'
+        path = f'{output_dir}/upset_plot.pdf'
         print("Saving upset plot to", path)
-        plt.savefig(path)
+        plt.savefig(path, bbox_inches='tight', format='pdf')

src/conformist/model_vs_cop_fnr.py

@@ -103,7 +103,7 @@ def run_reports(self):
         plt.legend(loc='upper right')
 
         # Save the plot to a file
-        fig.set_size_inches(4, 3)
+        fig.set_size_inches(4, 2)
         plt.tight_layout(w_pad=0)
         plt.savefig(f'{self.output_dir}/model_vs_CoP_FNR.pdf', format='pdf')
         print(f'Reports saved to {self.output_dir}')

src/conformist/performance_report.py

@@ -9,6 +9,10 @@ class PerformanceReport(OutputDir):
     FIGURE_WIDTH = 12
     FIGURE_HEIGHT = 8
     plt.rcParams.update({'font.size': FIGURE_FONTSIZE})
+    plt.rcParams["pdf.fonttype"] = 42  # If saving as PDF too
+    plt.rcParams["font.family"] = "Univers, DejaVu Sans"  # Choose a font that supports embedding
+
+
 
     def __init__(self, base_output_dir):
         self.create_output_dir(base_output_dir)
@@ -48,10 +52,12 @@ def _class_report(self,
                       items_by_class,
                       output_file_prefix,
                       ylabel,
-                      color):
+                      color,
+                      title=None):
         # Reset plt
-        plt.figure(figsize=(self.FIGURE_WIDTH,
-                            self.FIGURE_HEIGHT))
+        fig = plt.figure()
+        plt.tight_layout()
+        plt.rcParams.update({'font.size': 6})
 
         # Remove the grid
         plt.grid(False)
@@ -69,31 +75,27 @@ def _class_report(self,
 
         # Visualize this dict as a bar chart
         # sns.set_style('whitegrid')
-        fig, ax = plt.subplots(figsize=(
-            self.FIGURE_WIDTH,
-            self.FIGURE_HEIGHT))
+        fig, ax = plt.subplots()
+        plt.tight_layout()
         bars = ax.bar(range(len(mean_sizes)), mean_sizes.values(), color=color)
         ax.set_xticks(range(len(mean_sizes)))
         ax.set_xticklabels(mean_sizes.keys(),
                            rotation='vertical')
-        ax.tick_params(axis='both',
-                       labelsize=PerformanceReport.FIGURE_FONTSIZE)
-        ax.set_ylabel(ylabel,
-                      fontsize=PerformanceReport.FIGURE_FONTSIZE)
-        ax.set_xlabel('True class', fontsize=PerformanceReport.FIGURE_FONTSIZE)
+        ax.tick_params(axis='both')
+        ax.set_ylabel(ylabel)
+        ax.set_xlabel('True class')
 
         # Print the number above each bar
         for bar in bars:
             height = bar.get_height()
             ax.text(
                 bar.get_x() + bar.get_width() / 2.0, height, f'{height:.2f}',
                 ha='center',
-                va='bottom',
-                fontsize=PerformanceReport.FIGURE_FONTSIZE)
-
+                va='bottom')
 
-        plt.tight_layout()
-        plt.savefig(f'{self.output_dir}/{output_file_prefix}.png')
+        fig.set_size_inches(4, 3)
+        plt.tight_layout(w_pad=0)
+        plt.savefig(f'{self.output_dir}/{output_file_prefix}.pdf', format='pdf')
 
     def visualize_mean_set_sizes_by_class(self,
                                           mean_set_sizes_by_class):

src/conformist/prediction_dataset.py

@@ -660,7 +660,7 @@ def visualize_prediction_stripplot(self,
                             loc='lower center',
                             frameon=False,
                             ncol=4,
-                            bbox_to_anchor=(0.35, -0.2),
+                            bbox_to_anchor=(0.35, -0.3),
                             handletextpad=1,  # Increase padding between legend handle and text
                             columnspacing=2  # Increase spacing between columns
                             )
@@ -680,7 +680,7 @@ def visualize_prediction_stripplot(self,
 
         # plt.tight_layout()
 
-        fig.set_size_inches(3.75, 6)
+        fig.set_size_inches(4, 4)
 
         # Save the plot to a file
         # plt.subplots_adjust(hspace=0.5)