feat: Improved plots

results/make_graphs.py

@@ -1,3 +1,4 @@
+import glob
 import os
 
 import matplotlib.pyplot as plt
@@ -67,7 +68,32 @@ def compression_ratios(df: pd.DataFrame, unique_labels, palette_dict) -> Figure:
 
     plt.yticks(rotation=45, ha="right")
 
-    ax.grid(True)
+    ax.grid(True, alpha=0.3)
+
+    return plt.gcf()
+
+
+def compression_ratio_v_compression_time(df: pd.DataFrame, unique_labels, palette_dict, markers_dict) -> Figure:
+    plt.figure()
+
+    sns.scatterplot(
+        data=df,
+        x=RATE_COL,
+        y=COMPRESS_TIME_COL,
+        hue=LABEL_COL,
+        hue_order=unique_labels,
+        palette=palette_dict,
+        style=LABEL_COL,
+        style_order=unique_labels,
+        markers=markers_dict,
+    )
+
+    plt.legend(title='Compressor')
+
+    plt.xlabel('Compression ratio')
+    plt.ylabel('Compression time (s)')
+
+    plt.grid(True, alpha=0.3)
 
     return plt.gcf()
 
@@ -120,25 +146,35 @@ def filesize_v_mse(df: pd.DataFrame) -> Figure:
         size = row[INPUT_SIZE_COL]
         return f"{filename} ({size:.4f} MB)"
 
-    df['input_filename_size'] = df.apply(filename_and_size, axis=1)
+    def size(row):
+        full_name = row['input_filename']
+        # Strip prefix
+        size_name = full_name.lstrip('text').lstrip('genome').rstrip('txt').rstrip('fna')
+        size_name = f"*{size_name}*"
+        return size_name
 
-    fig, ax = plt.subplots()
+    df['input_filename_size'] = df.apply(filename_and_size, axis=1)
-    sns.barplot(
+    df['input_size'] = df.apply(size, axis=1)
+
+    g = sns.catplot(
         data=df,
-        y='input_filename',
+        kind="bar",
         x=DISTORTION_COL,
+        y='input_size',
+        col='training_dataset',
         hue=CONTEXT_COL,
-        ax=ax,
+        palette='Set2',
-        palette='Set2'
+        height=5,
+        aspect=0.6
     )
 
-    plt.title('MSE for autoencoder')
+    g.set_axis_labels("MSE Loss", "Filename")
-    plt.xlabel('MSE')
+    g.set_titles("Autoencoder trained on {col_name}")
-    plt.ylabel('Filename')
-    plt.yticks(rotation=45, ha="right")
-    plt.legend(title='Context size')
 
-    plt.grid(True)
+    # plt.title('MSE for autoencoder')
+    # plt.yticks(rotation=45, ha="right")
+    # plt.legend(title='Context size')
+    g.tight_layout()
 
     return plt.gcf()
 
@@ -164,7 +200,7 @@ def mse_losses(df: pd.DataFrame, unique_labels, palette_dict) -> Figure:
 
     plt.yticks(rotation=45, ha="right")
 
-    ax.grid(True)
+    ax.grid(True, alpha=0.3)
 
     return plt.gcf()
 
@@ -238,8 +274,8 @@ def split_graph(
     f.text(0.5, 0, x_axis_label, ha='center', va='center')
     ax_left.set_ylabel(y_axis_label)
 
-    ax_left.grid(True)
+    ax_left.grid(True, alpha=0.3)
-    ax_right.grid(True)
+    ax_right.grid(True, alpha=0.3)
 
     plt.tight_layout()
     return f, ax_left, ax_right
@@ -247,7 +283,7 @@ def split_graph(
 
 def generate(
         df: pd.DataFrame, unique_labels, palette_dict, markers_dict,
-        tgt_dir: str, dpi: int = 300
+        tgt_dir: str
 ) -> None:
     """Generate all the plots"""
     # Make plots
@@ -270,6 +306,11 @@ def generate(
         os.path.join(tgt_dir, 'compression_ratios.png'),
         bbox_inches='tight'
     )
+    compression_ratio_v_compression_time(df, unique_labels, palette_dict, markers_dict).savefig(
+        os.path.join(tgt_dir, 'compression_ratio_v_compression_time.png'),
+        bbox_inches='tight'
+    )
+
     filesize_v_mse(df).savefig(
         os.path.join(tgt_dir, 'filesize_mse.png'),
         bbox_inches='tight'
@@ -280,7 +321,7 @@ def generate(
     )
 
 
-def setup(tgt_dir):
+def setup(tgt_dir, dpi = 300):
     # Create the targ directory if it does not exist
     os.makedirs(tgt_dir, exist_ok=True)
 
@@ -288,7 +329,7 @@ def setup(tgt_dir):
     params = {'text.usetex': True,
               'font.size': 11,
               'font.family': 'serif',
-              'figure.dpi': 300,
+              'figure.dpi': dpi,
               }
     plt.rcParams.update(params)
 
@@ -327,8 +368,8 @@ def main():
     df = pd.read_csv("measurements.csv")
 
     tgt_dir = "figures"
-    setup(tgt_dir)
+    setup(tgt_dir, 300)
-    generate(*preprocessing(df), tgt_dir=tgt_dir, dpi=150)
+    generate(*preprocessing(df), tgt_dir=tgt_dir)
 
 
 def old_results():
@@ -395,7 +436,7 @@ def old_results():
             # plt.title(f"{model_type.capitalize()} compressed file evolution: {dataset_type}")
             plt.xlabel("Original file size (MB)")
             plt.ylabel("Compressed file size (MB)")
-            plt.ylim(0, model_df["compressed_file_size"].max() / 1e6)
+            plt.ylim(0, df[df["model_type"] == model_type]["compressed_file_size"].max() / 1e6)
             plt.legend(title="Context size")
             plt.tight_layout()
             plt.savefig(f"./graphs/{model_type}_{dataset_type}_compression_ratio.png")
@@ -428,7 +469,7 @@ def old_results():
                         linestyle=linestyle
                     )
 
-                plt.grid(True)
+                plt.grid(True, alpha=0.3)
                 plt.legend()
                 plt.title(f"(Log-linear) Extrapolated execution time for CNN")
                 # plt.xscale('log')
@@ -466,11 +507,43 @@ def old_results():
         plt.xlabel("MSE loss")
         plt.ylabel("Filename")
         plt.legend()
-        plt.grid(True)
+        plt.grid(True, alpha=0.3)
         plt.tight_layout()
         plt.savefig(f"./graphs/{model_type}_loss.png")
 
 
+def training_loss(df, loss) -> Figure:
+    plt.figure(figsize=(4, 3))
+
+    plt.plot(df['train_loss'], label="Training loss")
+    plt.plot(df['validation_loss'], label="Validation losses")
+
+    plt.xlabel("Epoch")
+    plt.ylabel(loss)
+
+    if loss == 'MSE Loss':
+        ylim = 0.01
+    else:
+        ylim = 6
+    plt.ylim(0, ylim)
+
+    plt.legend()
+    plt.tight_layout()
+
+    return plt.gcf()
+
+
+def make_training_graphs(models_dir, loss):
+    for csv in glob.glob(models_dir + '/*.csv'):
+        df = pd.read_csv(csv)
+        training_loss(df, loss).savefig(
+            csv.replace('.csv', '.png'),
+            bbox_inches='tight',
+        )
+
+
 if __name__ == "__main__":
     main()
-    old_results()
+    # old_results()
+    make_training_graphs('../models/autoencoder', 'MSE Loss')
+    make_training_graphs('../models/cnn', 'Cross Entropy Loss')