new file: checkpoints_tree/task1_tr_tree/best_tr_tree.pkl

new file: checkpoints_tree/task1_tr_tree/model_selection.csv new file: evaluation_outputs/task1_tr_tree/evaluation_train_all_samples.csv new file: evaluation_outputs/task1_tr_tree/evaluation_train_curve.png new file: evaluation_outputs/task1_tr_tree/evaluation_val_all_samples.csv new file: evaluation_outputs/task1_tr_tree/evaluation_val_curve.png new file: evaluation_outputs/task1_tr_tree/harmonic_5mm_1.55Hz_prediction.png new file: scripts_tree/__pycache__/config.cpython-310.pyc new file: scripts_tree/config.py new file: scripts_tree/evaluate.py new file: scripts_tree/predict_single.py new file: scripts_tree/train.py
2026-05-06 12:32:49 +08:00
parent 484643409d
commit 02e488cd0d
12 changed files with 549 additions and 0 deletions
--- a/scripts_tree/evaluate.py
+++ b/scripts_tree/evaluate.py
@@ -0,0 +1,177 @@
+from __future__ import annotations
+
+import argparse
+import pickle
+from pathlib import Path
+import sys
+
+import matplotlib.pyplot as plt
+import pandas as pd
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+if str(PROJECT_ROOT) not in sys.path:
+    sys.path.insert(0, str(PROJECT_ROOT))
+
+from scripts.dataset import build_datasets, report_to_text
+
+try:
+    from .config import FEATURE_NAMES, evaluation_dir, make_experiment_config
+except ImportError:
+    from config import FEATURE_NAMES, evaluation_dir, make_experiment_config
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Evaluate tree model on train/val/test splits.")
+    parser.add_argument("--split", choices=("train", "val", "test"), default="val")
+    parser.add_argument("--sample-index", type=int, default=0)
+    parser.add_argument("--all-samples", action="store_true")
+    parser.add_argument("--checkpoint", type=str, default=None)
+    return parser.parse_args()
+
+
+def resolve_checkpoint_path(project_root: Path, checkpoint_arg: str | None) -> Path:
+    if checkpoint_arg:
+        return Path(checkpoint_arg).resolve()
+    return project_root / "checkpoints_tree" / "task1_tr_tree" / "best_tr_tree.pkl"
+
+
+def relative_percent_error(true_value: float, pred_value: float) -> float:
+    return abs(pred_value - true_value) / max(abs(true_value), 1e-12) * 100.0
+
+
+def evaluate_record(model, record) -> dict[str, float | str]:
+    feature_matrix = [record.features.tolist()]
+    pred_tr = max(float(model.predict(feature_matrix)[0]), 1e-6)
+    x_rms = float(record.x_rms.item())
+    y_rms = float(record.y_rms.item())
+    pred_rms = pred_tr * x_rms
+    row = {
+        "file_name": record.file_path.name,
+        "frequency_hz": float(record.frequency_hz),
+        "true_rms": y_rms,
+        "pred_rms": pred_rms,
+        "relative_error_percent": relative_percent_error(y_rms, pred_rms),
+        "x_rms": x_rms,
+        "pred_tr": pred_tr,
+        "true_tr": float(record.target_y_rms.item()),
+    }
+    for name, value in zip(FEATURE_NAMES, record.features.tolist()):
+        row[name] = float(value)
+    row["sampling_rate"] = float(record.sampling_rate)
+    return row
+
+
+def save_all_samples_plot(result_df: pd.DataFrame, split: str, save_dir: Path) -> Path | None:
+    if result_df["frequency_hz"].isna().any():
+        return None
+    plot_df = result_df.sort_values("frequency_hz").reset_index(drop=True)
+    fig, axes = plt.subplots(2, 1, figsize=(12, 8))
+    fig.suptitle(f"Task1 TR Tree Evaluation | {split}")
+
+    axes[0].plot(plot_df["frequency_hz"], plot_df["true_rms"], marker="o", label="True RMS")
+    axes[0].plot(plot_df["frequency_hz"], plot_df["pred_rms"], marker="o", label="Pred RMS")
+    axes[0].set_xlabel("Frequency (Hz)")
+    axes[0].set_ylabel("RMS")
+    axes[0].grid(True, alpha=0.3)
+    axes[0].legend()
+
+    axes[1].bar(plot_df["frequency_hz"].astype(str), plot_df["relative_error_percent"], color="tab:orange")
+    axes[1].set_xlabel("Frequency (Hz)")
+    axes[1].set_ylabel("Relative Error (%)")
+    axes[1].grid(True, axis="y", alpha=0.3)
+    axes[1].tick_params(axis="x", labelrotation=45)
+
+    plt.tight_layout()
+    figure_path = save_dir / f"evaluation_{split}_curve.png"
+    plt.savefig(figure_path, dpi=180, bbox_inches="tight")
+    plt.close(fig)
+    return figure_path
+
+
+def save_single_sample_plot(record, result: dict[str, float | str], split: str, save_dir: Path, sample_index: int) -> Path:
+    time_middle = record.time_middle.detach().cpu().numpy().reshape(-1)
+    x_middle = record.x_middle.detach().cpu().numpy().reshape(-1)
+    y_middle = record.y_middle.detach().cpu().numpy().reshape(-1)
+    frequency_hz = float(result["frequency_hz"])
+    pred_rms = float(result["pred_rms"])
+    true_rms = float(result["true_rms"])
+
+    fig, axes = plt.subplots(3, 1, figsize=(12, 10))
+    fig.suptitle(f"Task1 TR Tree | {split} | {result['file_name']}")
+
+    axes[0].plot(time_middle, x_middle, color="tab:blue")
+    axes[0].set_title("Input Base Excitation (Middle Segment)")
+    axes[0].set_xlabel("Time")
+    axes[0].set_ylabel("Acceleration")
+    axes[0].grid(True, alpha=0.3)
+
+    axes[1].plot(time_middle, y_middle, color="tab:green")
+    axes[1].set_title("True Top Response (Middle Segment)")
+    axes[1].set_xlabel("Time")
+    axes[1].set_ylabel("Acceleration")
+    axes[1].grid(True, alpha=0.3)
+
+    axes[2].bar(["True RMS", "Pred RMS"], [true_rms, pred_rms], color=["tab:green", "tab:orange"])
+    axes[2].set_title(
+        f"Freq: {frequency_hz:.4f} Hz | True RMS: {true_rms:.6f} | Pred RMS: {pred_rms:.6f} | "
+        f"Error: {float(result['relative_error_percent']):.2f}%"
+    )
+    axes[2].set_ylabel("RMS")
+    axes[2].grid(True, axis="y", alpha=0.3)
+
+    plt.tight_layout()
+    figure_path = save_dir / f"evaluation_{split}_s{sample_index}.png"
+    plt.savefig(figure_path, dpi=180, bbox_inches="tight")
+    plt.close(fig)
+    return figure_path
+
+
+def main() -> None:
+    args = parse_args()
+    config = make_experiment_config()
+    ckpt_path = resolve_checkpoint_path(config.data.project_root, args.checkpoint)
+    with ckpt_path.open("rb") as handle:
+        bundle = pickle.load(handle)
+    model = bundle["model"]
+
+    _, raw_records, reports = build_datasets(config.data)
+    print(report_to_text(reports))
+    records = raw_records[args.split]
+    save_dir = evaluation_dir(config)
+
+    if args.all_samples:
+        rows = [evaluate_record(model, record) for record in records]
+        result_df = pd.DataFrame(rows).sort_values(["relative_error_percent", "file_name"]).reset_index(drop=True)
+        csv_path = save_dir / f"evaluation_{args.split}_all_samples.csv"
+        result_df.to_csv(csv_path, index=False)
+        fig_path = save_all_samples_plot(result_df, args.split, save_dir)
+        summary = {
+            "count": len(result_df),
+            "mean_error_percent": float(result_df["relative_error_percent"].mean()),
+            "median_error_percent": float(result_df["relative_error_percent"].median()),
+            "max_error_percent": float(result_df["relative_error_percent"].max()),
+            "min_error_percent": float(result_df["relative_error_percent"].min()),
+        }
+        print(f"Checkpoint: {ckpt_path}")
+        print(f"Model: {bundle['model_name']}")
+        print(f"Summary: {summary}")
+        print(f"CSV saved to: {csv_path}")
+        if fig_path is not None:
+            print(f"Figure saved to: {fig_path}")
+        print(result_df.to_string(index=False))
+        return
+
+    record = records[args.sample_index]
+    result = evaluate_record(model, record)
+    fig_path = save_single_sample_plot(record, result, args.split, save_dir, args.sample_index)
+    print(f"Checkpoint: {ckpt_path}")
+    print(f"Model: {bundle['model_name']}")
+    print(f"Sample file: {result['file_name']}")
+    print(f"True RMS: {float(result['true_rms']):.6f}")
+    print(f"Pred RMS: {float(result['pred_rms']):.6f}")
+    print(f"Relative RMS Error (%): {float(result['relative_error_percent']):.4f}")
+    print(f"Figure saved to: {fig_path}")
+
+
+if __name__ == "__main__":
+    main()