refactor configuration files and update paths

analysis/ablation_lollipop.py

@@ -1,3 +1,4 @@
+import argparse
 import os
 import pandas as pd
 import numpy as np
@@ -15,7 +16,7 @@ def setup_barlow_font():
             rcParams['font.family'] = 'Barlow'
         else:
             for path in ['/usr/share/fonts/truetype/barlow/Barlow-Regular.ttf',
-                         '/Users/griffingoodwin/Library/Fonts/Barlow-Regular.otf']:
+                         os.path.expanduser('~/Library/Fonts/Barlow-Regular.otf')]:
                 if os.path.exists(path):
                     fm.fontManager.addfont(path)
                     rcParams['font.family'] = 'Barlow'
@@ -27,9 +28,6 @@ def setup_barlow_font():
 
 setup_barlow_font()
 
-DATA_DIR     = "/Users/griffingoodwin/Documents/gitrepos/FOXES/Untracked/data"
-BASELINE_CSV = "/Volumes/T9/FOXES_Misc/batch_results/vit/vit_predictions_test.csv"
-
 WAVELENGTHS = ["94", "131", "171", "193", "211", "304", "335","STEREO"]
 LABELS = {
     "94":     "Ablate 94 Å",
@@ -67,96 +65,110 @@ def compute_row(label, gt, pred, is_baseline=False):
         row[cls] = np.mean(np.abs(np.log10(gt[m]) - np.log10(pred[m]))) if m.sum() > 5 else np.nan
     return row
 
-records = []
-
-# Baseline
-bl = pd.read_csv(BASELINE_CSV)
-records.append(compute_row("FOXES (no ablation)",
-                           bl["groundtruth"].values, bl["predictions"].values,
-                           is_baseline=True))
-
-for wav in WAVELENGTHS:
-    ab = pd.read_csv(f"{DATA_DIR}/ablate_{wav}_global_1.csv")
-    records.append(compute_row(LABELS[wav], ab["groundtruth"].values, ab["predictions"].values))
-
-# Sort ablation rows by overall MAE (worst first), keep baseline pinned at bottom
-ablation_df = pd.DataFrame([r for r in records if not r["is_baseline"]])
-ablation_df = ablation_df.sort_values("overall", ascending=False).reset_index(drop=True)
-baseline_df = pd.DataFrame([r for r in records if r["is_baseline"]])
-df = pd.concat([ablation_df, baseline_df], ignore_index=True)
-
-# ── Plot ───────────────────────────────────────────────────────────────────────
-n_rows = len(df)
-fig, ax = plt.subplots(figsize=(11, 0.6 * n_rows + 1.5))
-#ax.set_facecolor("#FAFAFA")
-fig.patch.set_facecolor("#FFFFFF")
-
-y_positions = np.arange(n_rows)
-
-# Separator line between ablations and baseline
-ax.axhline(y=n_rows - 1.5, color="#BBBBBB", linewidth=1, linestyle=":", zorder=1)
-
-for i, row in df.iterrows():
-    y = y_positions[i]
-    is_bl = row["is_baseline"]
-
-    # Highlight baseline row
-    if is_bl:
-        ax.axhspan(y - 0.45, y + 0.45, color="#EEF6FF", zorder=0)
-
-    # Span line across per-class range
-    class_vals = [row[c] for c in FLARE_CLASSES if not np.isnan(row[c])]
-    if class_vals:
-        ax.hlines(y, min(class_vals), max(class_vals),
-                  color="#CCCCCC", linewidth=2, zorder=1)
-
-    # Stem from 0 to overall
-    ax.hlines(y, 0, row["overall"],
-              color="#AAAAAA", linewidth=1.2, linestyle="--", zorder=0, alpha=0.6)
-
-    # Per-class dots
-    for cls in FLARE_CLASSES:
-        val = row[cls]
-        if not np.isnan(val):
-            ax.scatter(val, y, color=CLASS_COLORS[cls], s=80, zorder=4,
-                       edgecolors="white", linewidths=0.6, alpha=0.75)
-
-    # Overall dot
-    outline_color = "#1A6BBF" if is_bl else "black"
-    ax.scatter(row["overall"], y, color="white", s=190, zorder=3,
-               edgecolors=outline_color, linewidths=2.0 if is_bl else 1.5, alpha=0.75)
-    ax.scatter(row["overall"], y, color=outline_color, s=75, zorder=3,
-               marker="|", linewidths=1.5, alpha=0.75)
-
-tick_colors = ["black"] * n_rows
-tick_colors[-1] = "#1A6BBF"  # baseline label in blue
-ax.set_yticks(y_positions)
-ax.set_yticklabels(df["label"], fontsize=12)
-for ticklabel, color in zip(ax.get_yticklabels(), tick_colors):
-    ticklabel.set_color(color)
-    if color != "black":
-        ticklabel.set_fontweight("bold")
-ax.set_xlabel("MAE (log$_{10}$ scale)", fontsize=12)
-ax.grid(True, axis="x", alpha=0.4, color="#CCCCCC", linewidth=0.6)
-ax.set_axisbelow(True)
-ax.spines[["top", "right"]].set_visible(False)
-ax.tick_params(axis="y", length=0, labelsize=11)
-ax.tick_params(axis="x", labelsize=10)
-
-# Legend
-class_patches = [
-    mpatches.Patch(color=CLASS_COLORS[c], label=f"{c}-class") for c in FLARE_CLASSES
-]
-overall_patch   = mpatches.Patch(facecolor="white", edgecolor="black",  label="Overall")
-#baseline_patch  = mpatches.Patch(facecolor="white", edgecolor="#1A6BBF", label="Baseline (overall)")
-ax.legend(handles=class_patches + [overall_patch],
-          loc="upper right", fontsize=10, framealpha=0.9,
-          edgecolor="#CCCCCC")
-
-# ax.set_title("Ablation Study — Log MAE by Channel & Flare Class",
-#              fontsize=14, fontweight="bold", pad=14)
-plt.xlim(0, .85)
-plt.tight_layout()
-plt.savefig("ablation_lollipop.png", dpi=450, bbox_inches="tight")
-plt.show()
-print("Saved: analysis/ablation_lollipop.png")
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Ablation lollipop plot")
+    parser.add_argument("--data_dir",      required=True,
+                        help="Directory containing ablate_<wavelength>_global_1.csv files")
+    parser.add_argument("--baseline_csv",  required=True,
+                        help="Baseline predictions CSV (groundtruth + predictions columns)")
+    parser.add_argument("--out",           default="ablation_lollipop.png",
+                        help="Output image path (default: ablation_lollipop.png)")
+    args = parser.parse_args()
+
+    DATA_DIR     = args.data_dir
+    BASELINE_CSV = args.baseline_csv
+    OUT_PATH     = args.out
+
+    records = []
+
+    # Baseline
+    bl = pd.read_csv(BASELINE_CSV)
+    records.append(compute_row("FOXES (no ablation)",
+                               bl["groundtruth"].values, bl["predictions"].values,
+                               is_baseline=True))
+
+    for wav in WAVELENGTHS:
+        ab = pd.read_csv(f"{DATA_DIR}/ablate_{wav}_global_1.csv")
+        records.append(compute_row(LABELS[wav], ab["groundtruth"].values, ab["predictions"].values))
+
+    # Sort ablation rows by overall MAE (worst first), keep baseline pinned at bottom
+    ablation_df = pd.DataFrame([r for r in records if not r["is_baseline"]])
+    ablation_df = ablation_df.sort_values("overall", ascending=False).reset_index(drop=True)
+    baseline_df = pd.DataFrame([r for r in records if r["is_baseline"]])
+    df = pd.concat([ablation_df, baseline_df], ignore_index=True)
+
+    # ── Plot ───────────────────────────────────────────────────────────────────────
+    n_rows = len(df)
+    fig, ax = plt.subplots(figsize=(11, 0.6 * n_rows + 1.5))
+    #ax.set_facecolor("#FAFAFA")
+    fig.patch.set_facecolor("#FFFFFF")
+
+    y_positions = np.arange(n_rows)
+
+    # Separator line between ablations and baseline
+    ax.axhline(y=n_rows - 1.5, color="#BBBBBB", linewidth=1, linestyle=":", zorder=1)
+
+    for i, row in df.iterrows():
+        y = y_positions[i]
+        is_bl = row["is_baseline"]
+
+        # Highlight baseline row
+        if is_bl:
+            ax.axhspan(y - 0.45, y + 0.45, color="#EEF6FF", zorder=0)
+
+        # Span line across per-class range
+        class_vals = [row[c] for c in FLARE_CLASSES if not np.isnan(row[c])]
+        if class_vals:
+            ax.hlines(y, min(class_vals), max(class_vals),
+                      color="#CCCCCC", linewidth=2, zorder=1)
+
+        # Stem from 0 to overall
+        ax.hlines(y, 0, row["overall"],
+                  color="#AAAAAA", linewidth=1.2, linestyle="--", zorder=0, alpha=0.6)
+
+        # Per-class dots
+        for cls in FLARE_CLASSES:
+            val = row[cls]
+            if not np.isnan(val):
+                ax.scatter(val, y, color=CLASS_COLORS[cls], s=80, zorder=4,
+                           edgecolors="white", linewidths=0.6, alpha=0.75)
+
+        # Overall dot
+        outline_color = "#1A6BBF" if is_bl else "black"
+        ax.scatter(row["overall"], y, color="white", s=190, zorder=3,
+                   edgecolors=outline_color, linewidths=2.0 if is_bl else 1.5, alpha=0.75)
+        ax.scatter(row["overall"], y, color=outline_color, s=75, zorder=3,
+                   marker="|", linewidths=1.5, alpha=0.75)
+
+    tick_colors = ["black"] * n_rows
+    tick_colors[-1] = "#1A6BBF"  # baseline label in blue
+    ax.set_yticks(y_positions)
+    ax.set_yticklabels(df["label"], fontsize=12)
+    for ticklabel, color in zip(ax.get_yticklabels(), tick_colors):
+        ticklabel.set_color(color)
+        if color != "black":
+            ticklabel.set_fontweight("bold")
+    ax.set_xlabel("MAE (log$_{10}$ scale)", fontsize=12)
+    ax.grid(True, axis="x", alpha=0.4, color="#CCCCCC", linewidth=0.6)
+    ax.set_axisbelow(True)
+    ax.spines[["top", "right"]].set_visible(False)
+    ax.tick_params(axis="y", length=0, labelsize=11)
+    ax.tick_params(axis="x", labelsize=10)
+
+    # Legend
+    class_patches = [
+        mpatches.Patch(color=CLASS_COLORS[c], label=f"{c}-class") for c in FLARE_CLASSES
+    ]
+    overall_patch   = mpatches.Patch(facecolor="white", edgecolor="black",  label="Overall")
+    #baseline_patch  = mpatches.Patch(facecolor="white", edgecolor="#1A6BBF", label="Baseline (overall)")
+    ax.legend(handles=class_patches + [overall_patch],
+              loc="upper right", fontsize=10, framealpha=0.9,
+              edgecolor="#CCCCCC")
+
+    # ax.set_title("Ablation Study — Log MAE by Channel & Flare Class",
+    #              fontsize=14, fontweight="bold", pad=14)
+    plt.xlim(0, .85)
+    plt.tight_layout()
+    plt.savefig(OUT_PATH, dpi=450, bbox_inches="tight")
+    plt.show()
+    print(f"Saved: {OUT_PATH}")

analysis/spatial_performance.py

@@ -35,10 +35,10 @@ sys.path.insert(0, str(PROJECT_ROOT))
 from forecasting.inference.evaluation import setup_barlow_font
 
 # ---------------------------------------------------------------------------
-# Paths — edit here
+# Paths — override via CLI args or environment variables
 # ---------------------------------------------------------------------------
-FLUX_DIR        = "/Volumes/T9/FOXES_Data/flux/"
-PREDICTIONS_CSV = "/Volumes/T9/FOXES_Misc/batch_results/vit/vit_predictions_test.csv"
+FLUX_DIR        = os.environ.get("FOXES_FLUX_DIR", "")
+PREDICTIONS_CSV = os.environ.get("FOXES_PREDICTIONS_CSV", "")
 OUT_DIR         = Path(__file__).parent
 GRID_SIZE       = 64    # 512px / 8px patch size
 BIN_SIZE        = 1    # downsample factor (1 = full 64×64 resolution)

forecasting/inference/ablation_inference_config.yaml

@@ -11,7 +11,7 @@
 # Define top-level string keys and reference them anywhere with ${key}.
 
 base_dir:   "/Volumes/T9/FOXES_Data"
-checkpoint: "/Users/griffingoodwin/Downloads/FOXES_Model_Checkpoint.ckpt"
+checkpoint: ""  # Path to your model checkpoint (.ckpt)
 
 model: "ViTLocal"
 wavelengths: [94, 131, 171, 193, 211, 304, 335]

forecasting/inference/evaluation.py

@@ -39,7 +39,7 @@ def setup_barlow_font():
         else:
             # Try alternative approach - directly specify font file
             barlow_path = '/usr/share/fonts/truetype/barlow/Barlow-Regular.ttf'
-            barlow_path2 = '/Users/griffingoodwin/Library/Fonts/Barlow-Regular.otf'
+            barlow_path2 = os.path.expanduser('~/Library/Fonts/Barlow-Regular.otf')
             if os.path.exists(barlow_path):
                 # Add the font file directly to matplotlib
                 fm.fontManager.addfont(barlow_path)

forecasting/training/train_config.yaml

@@ -44,12 +44,12 @@ data:
     "${base_checkpoint_dir}/new-checkpoint/"
 
 wandb:
-  entity: jayantbiradar619-university-of-arizona # Use your exact W&B username
-  project: Paper
+  entity: ""  # Set to your W&B username or team name
+  project: FOXES
   job_type: training
   tags:
     - aia
     - sxr
     - regression
-  run_name: paper-8-patch-4ch
-  notes: Regression from AIA images to SXR images using ViTLocal model with 8x8 patches
+  run_name: run_1
+  notes: AIA to SXR translation

pipeline_config.yaml

@@ -14,7 +14,7 @@
 # Change base_dir or checkpoint once and every path updates automatically.
 
 base_dir:   "/Volumes/T9/FOXES_Data"
-checkpoint: "/Users/griffingoodwin/Downloads/FOXES_Model_Checkpoint.ckpt"
+checkpoint: ""  # Path to your model checkpoint (.ckpt)
 
 # -----------------------------------------------------------------------------
 # HuggingFace download  (step: hf_download)
@@ -102,7 +102,7 @@ train:
     batch_size: 6
     wandb:
       run_name: "pipeline-run"
-      entity: jayantbiradar619-university-of-arizona
+      entity: ""  # Set to your W&B username or team name
       project: Paper
       job_type: training
       tags: