File size: 28,264 Bytes
0a6452f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 |
#!/usr/bin/env python3
"""
评估脚本
Evaluation script for PAD Predictor
该脚本实现了完整的模型评估流程,包括:
- 加载训练好的模型
- 测试集评估和性能分析
- 生成详细的评估报告和可视化图表
- 支持模型比较和批量评估
- 置信度校准分析
- PAD特定指标分析
使用方法:
python evaluate.py --model-path checkpoints/best_model.pth --data-path data/test.csv
python evaluate.py --model-path checkpoints/final_model.pth --config configs/training_config.yaml
python evaluate.py --compare-models model1.pth model2.pth --data-path data/test.csv
"""
import argparse
import os
import sys
import yaml
import json
import torch
import torch.nn as nn
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from pathlib import Path
from typing import Dict, List, Any, Optional, Union, Tuple
import logging
import warnings
from datetime import datetime
from collections import defaultdict
# 添加项目根目录到Python路径
project_root = Path(__file__).parent.parent.parent
sys.path.insert(0, str(project_root))
from src.models.pad_predictor import PADPredictor, create_pad_predictor
from src.data.data_loader import DataLoader, load_data_from_config
from src.models.metrics import PADMetrics, RegressionMetrics, CalibrationMetrics
from src.utils.logger import TrainingLogger, create_logger
def parse_arguments() -> argparse.Namespace:
"""
解析命令行参数
Returns:
解析后的参数
"""
parser = argparse.ArgumentParser(
description='PAD预测器评估脚本',
formatter_class=argparse.ArgumentDefaultsHelpFormatter
)
# 模型参数
parser.add_argument(
'--model-path', '-m',
type=str,
required=True,
help='模型文件路径'
)
parser.add_argument(
'--model-config', '-mc',
type=str,
default='configs/model_config.yaml',
help='模型配置文件路径'
)
# 数据参数
parser.add_argument(
'--data-path', '-d',
type=str,
help='测试数据文件路径'
)
parser.add_argument(
'--config', '-c',
type=str,
default='configs/training_config.yaml',
help='训练配置文件路径(用于数据加载配置)'
)
# 输出参数
parser.add_argument(
'--output-dir', '-o',
type=str,
default='evaluation_results',
help='输出目录'
)
parser.add_argument(
'--report-name', '-r',
type=str,
default='evaluation_report',
help='评估报告名称'
)
# 评估参数
parser.add_argument(
'--batch-size', '-b',
type=int,
help='批次大小(覆盖配置文件中的设置)'
)
parser.add_argument(
'--device',
type=str,
choices=['auto', 'cpu', 'cuda', 'mps'],
default='auto',
help='评估设备'
)
parser.add_argument(
'--gpu-id',
type=int,
default=0,
help='GPU ID(当使用CUDA时)'
)
# 比较评估参数
parser.add_argument(
'--compare-models',
nargs='+',
type=str,
help='比较多个模型,提供模型路径列表'
)
parser.add_argument(
'--model-names',
nargs='+',
type=str,
help='比较模型时的名称列表'
)
# 分析参数
parser.add_argument(
'--detailed-analysis',
action='store_true',
help='进行详细分析(包括组件级别分析)'
)
parser.add_argument(
'--calibration-analysis',
action='store_true',
help='进行置信度校准分析'
)
parser.add_argument(
'--error-analysis',
action='store_true',
help='进行误差分析'
)
parser.add_argument(
'--generate-plots',
action='store_true',
default=True,
help='生成可视化图表'
)
# 数据参数
parser.add_argument(
'--synthetic-data',
action='store_true',
help='使用合成数据进行评估'
)
parser.add_argument(
'--num-samples',
type=int,
default=1000,
help='合成数据样本数量'
)
# 其他参数
parser.add_argument(
'--verbose', '-v',
action='store_true',
help='详细输出'
)
parser.add_argument(
'--save-predictions',
action='store_true',
help='保存预测结果'
)
parser.add_argument(
'--format',
choices=['json', 'csv', 'xlsx'],
default='json',
help='输出格式'
)
return parser.parse_args()
def load_model(model_path: str,
model_config: Optional[Dict[str, Any]] = None,
device: Union[str, torch.device] = 'cpu') -> nn.Module:
"""
加载模型
Args:
model_path: 模型文件路径
model_config: 模型配置
device: 设备
Returns:
加载的模型
"""
if not os.path.exists(model_path):
raise FileNotFoundError(f"模型文件不存在: {model_path}")
# 加载检查点
checkpoint = torch.load(model_path, map_location=device)
# 从检查点获取模型配置
if model_config is None and 'model_config' in checkpoint:
model_config = checkpoint['model_config']
elif model_config is None:
# 使用默认配置
model_config = {
'dimensions': {'input_dim': 10, 'output_dim': 4}, # 10维输入,4维输出(移除confidence)
'architecture': {
'hidden_layers': [
{'size': 128, 'activation': 'ReLU', 'dropout': 0.2},
{'size': 64, 'activation': 'ReLU', 'dropout': 0.2},
{'size': 32, 'activation': 'ReLU', 'dropout': 0.1}
]
},
'initialization': {'weight_init': 'xavier_uniform', 'bias_init': 'zeros'}
}
# 创建模型
model = create_pad_predictor(model_config)
# 加载权重
if 'model_state_dict' in checkpoint:
model.load_state_dict(checkpoint['model_state_dict'])
else:
model.load_state_dict(checkpoint)
model.to(device)
model.eval()
logging.info(f"模型已加载: {model_path}")
logging.info(f"模型参数数量: {sum(p.numel() for p in model.parameters()):,}")
return model
def load_data_for_evaluation(config: Dict[str, Any],
data_path: Optional[str] = None,
synthetic_data: bool = False,
num_samples: int = 1000,
batch_size: Optional[int] = None) -> torch.utils.data.DataLoader:
"""
加载评估数据
Args:
config: 配置字典
data_path: 数据文件路径
synthetic_data: 是否使用合成数据
num_samples: 合成数据样本数量
batch_size: 批次大小
Returns:
数据加载器
"""
if synthetic_data:
# 使用合成数据
logging.info(f"生成合成数据,样本数量: {num_samples}")
from src.data.synthetic_generator import SyntheticDataGenerator
generator = SyntheticDataGenerator(num_samples=num_samples)
data, labels = generator.generate_data()
# 创建数据加载器
data_loader_config = config.get('data', {}).get('dataloader', {})
if batch_size:
data_loader_config['batch_size'] = batch_size
data_loader = DataLoader(data_loader_config)
test_loader = data_loader.get_test_loader(data=np.hstack([data, labels]))
else:
# 使用真实数据
if data_path:
# 从指定路径加载
logging.info(f"从文件加载数据: {data_path}")
data_loader_config = config.get('data', {}).get('dataloader', {})
if batch_size:
data_loader_config['batch_size'] = batch_size
data_loader = DataLoader(data_loader_config)
test_loader = data_loader.get_test_loader(data_path=data_path)
else:
# 从配置文件加载
logging.info("从配置文件加载数据")
_, _, test_loader = load_data_from_config(config.get('data', {}).get('test_data_path', ''))
logging.info(f"测试数据批次数: {len(test_loader)}")
return test_loader
def evaluate_model(model: nn.Module,
data_loader: torch.utils.data.DataLoader,
device: torch.device,
save_predictions: bool = False,
output_dir: Optional[str] = None) -> Dict[str, Any]:
"""
评估单个模型
Args:
model: 模型
data_loader: 数据加载器
device: 设备
save_predictions: 是否保存预测结果
output_dir: 输出目录
Returns:
评估结果
"""
model.eval()
all_predictions = []
all_targets = []
all_features = []
with torch.no_grad():
for features, targets in data_loader:
features = features.to(device)
targets = targets.to(device)
predictions = model(features)
all_predictions.append(predictions.cpu())
all_targets.append(targets.cpu())
all_features.append(features.cpu())
# 合并所有结果
all_predictions = torch.cat(all_predictions, dim=0)
all_targets = torch.cat(all_targets, dim=0)
all_features = torch.cat(all_features, dim=0)
# 计算评估指标
metrics = PADMetrics()
evaluation_results = metrics.evaluate_predictions(all_predictions, all_targets)
# 添加预测和目标数据
evaluation_results['predictions'] = all_predictions
evaluation_results['targets'] = all_targets
evaluation_results['features'] = all_features
# 保存预测结果
if save_predictions and output_dir:
predictions_file = Path(output_dir) / 'predictions.csv'
# 转换为DataFrame
pred_df = pd.DataFrame(all_predictions.numpy(),
columns=['delta_pad_p', 'delta_pad_a', 'delta_pad_d', 'delta_pressure', 'confidence'])
target_df = pd.DataFrame(all_targets.numpy(),
columns=['delta_pad_p', 'delta_pad_a', 'delta_pad_d', 'delta_pressure', 'confidence'])
feature_df = pd.DataFrame(all_features.numpy(),
columns=['user_pad_p', 'user_pad_a', 'user_pad_d', 'vitality', 'current_pad_p', 'current_pad_a', 'current_pad_d'])
# 合并数据
combined_df = pd.concat([feature_df, target_df, pred_df], axis=1)
combined_df.to_csv(predictions_file, index=False)
logging.info(f"预测结果已保存: {predictions_file}")
return evaluation_results
def generate_evaluation_report(results: Dict[str, Any],
output_dir: str,
report_name: str = 'evaluation_report',
detailed_analysis: bool = False,
calibration_analysis: bool = False,
error_analysis: bool = False,
generate_plots: bool = True) -> str:
"""
生成评估报告
Args:
results: 评估结果
output_dir: 输出目录
report_name: 报告名称
detailed_analysis: 是否进行详细分析
calibration_analysis: 是否进行校准分析
error_analysis: 是否进行误差分析
generate_plots: 是否生成图表
Returns:
报告文件路径
"""
output_path = Path(output_dir)
output_path.mkdir(parents=True, exist_ok=True)
# 生成文本报告
metrics = PADMetrics()
report_text = metrics.generate_evaluation_report(
results['predictions'],
results['targets'],
save_path=output_path / f'{report_name}.txt'
)
# 生成JSON报告
json_results = {}
for key, value in results.items():
if isinstance(value, torch.Tensor):
json_results[key] = value.tolist()
elif isinstance(value, dict):
json_results[key] = value
else:
json_results[key] = value
# 移除大型张量以减少文件大小
for key in ['predictions', 'targets', 'features']:
if key in json_results:
del json_results[key]
with open(output_path / f'{report_name}.json', 'w', encoding='utf-8') as f:
json.dump(json_results, f, indent=2, ensure_ascii=False)
if generate_plots:
generate_evaluation_plots(results, output_path, detailed_analysis, calibration_analysis, error_analysis)
logging.info(f"评估报告已生成: {output_path / f'{report_name}.txt'}")
return str(output_path / f'{report_name}.txt')
def generate_evaluation_plots(results: Dict[str, Any],
output_path: Path,
detailed_analysis: bool = False,
calibration_analysis: bool = False,
error_analysis: bool = False):
"""
生成评估图表
Args:
results: 评估结果
output_path: 输出路径
detailed_analysis: 是否进行详细分析
calibration_analysis: 是否进行校准分析
error_analysis: 是否进行误差分析
"""
predictions = results['predictions']
targets = results['targets']
# 设置图表样式
plt.style.use('seaborn-v0_8')
# 1. 预测vs真实值散点图
fig, axes = plt.subplots(2, 2, figsize=(14, 12)) # 改为 2x2 布局
fig.suptitle('预测值 vs 真实值', fontsize=16)
component_names = ['ΔPAD_P', 'ΔPAD_A', 'ΔPAD_D', 'ΔPressure'] # 4维输出(移除Confidence)
for i, (ax, name) in enumerate(zip(axes.flat, component_names)):
if i < predictions.size(1):
pred_vals = predictions[:, i].numpy()
true_vals = targets[:, i].numpy()
ax.scatter(true_vals, pred_vals, alpha=0.6, s=20)
# 添加对角线
min_val = min(true_vals.min(), pred_vals.min())
max_val = max(true_vals.max(), pred_vals.max())
ax.plot([min_val, max_val], [min_val, max_val], 'r--', linewidth=2)
ax.set_xlabel('真实值')
ax.set_ylabel('预测值')
ax.set_title(name)
ax.grid(True, alpha=0.3)
# 计算R²
r2 = np.corrcoef(true_vals, pred_vals)[0, 1] ** 2
ax.text(0.05, 0.95, f'R² = {r2:.3f}', transform=ax.transAxes,
bbox=dict(boxstyle='round', facecolor='white', alpha=0.8))
plt.tight_layout()
plt.savefig(output_path / 'prediction_vs_true.png', dpi=300, bbox_inches='tight')
plt.close()
# 2. 误差分布图
if detailed_analysis:
fig, axes = plt.subplots(2, 3, figsize=(18, 12))
fig.suptitle('误差分布', fontsize=16)
for i, (ax, name) in enumerate(zip(axes.flat, component_names)):
if i < predictions.size(1):
errors = (predictions[:, i] - targets[:, i]).numpy()
ax.hist(errors, bins=30, alpha=0.7, density=True)
ax.axvline(0, color='r', linestyle='--', linewidth=2)
ax.axvline(np.mean(errors), color='g', linestyle='-', linewidth=2, label=f'均值: {np.mean(errors):.4f}')
ax.axvline(np.median(errors), color='b', linestyle='-', linewidth=2, label=f'中位数: {np.median(errors):.4f}')
ax.set_xlabel('误差')
ax.set_ylabel('密度')
ax.set_title(name)
ax.legend()
ax.grid(True, alpha=0.3)
plt.tight_layout()
plt.savefig(output_path / 'error_distribution.png', dpi=300, bbox_inches='tight')
plt.close()
# 3. 校准分析图(已移除 - Confidence 不再作为输出维度)
# 注:置信度现在通过 MC Dropout 动态计算,不包含在模型输出中
# if calibration_analysis:
# # 如需校准分析,请使用 MC Dropout 获取预测置信度后再评估
# pass
# 4. PAD空间分析
if detailed_analysis:
# PAD向量角度分析
delta_pad_pred = predictions[:, :3]
delta_pad_true = targets[:, :3]
# 计算角度误差
cos_sim = torch.nn.functional.cosine_similarity(delta_pad_pred, delta_pad_true, dim=1)
angle_errors = torch.acos(torch.clamp(cos_sim, -1 + 1e-8, 1 - 1e-8)) * 180 / np.pi
fig, axes = plt.subplots(1, 2, figsize=(15, 6))
# 角度误差分布
axes[0].hist(angle_errors.numpy(), bins=30, alpha=0.7, density=True)
axes[0].set_xlabel('角度误差 (度)')
axes[0].set_ylabel('密度')
axes[0].set_title('PAD向量角度误差分布')
axes[0].grid(True, alpha=0.3)
# 余弦相似度分布
axes[1].hist(cos_sim.numpy(), bins=30, alpha=0.7, density=True)
axes[1].set_xlabel('余弦相似度')
axes[1].set_ylabel('密度')
axes[1].set_title('PAD向量余弦相似度分布')
axes[1].grid(True, alpha=0.3)
plt.tight_layout()
plt.savefig(output_path / 'pad_analysis.png', dpi=300, bbox_inches='tight')
plt.close()
logging.info(f"评估图表已保存到: {output_path}")
def compare_models(model_paths: List[str],
model_names: List[str],
data_loader: torch.utils.data.DataLoader,
device: torch.device,
output_dir: str) -> Dict[str, Any]:
"""
比较多个模型
Args:
model_paths: 模型路径列表
model_names: 模型名称列表
data_loader: 数据加载器
device: 设备
output_dir: 输出目录
Returns:
比较结果
"""
if len(model_names) != len(model_paths):
model_names = [f"Model_{i+1}" for i in range(len(model_paths))]
comparison_results = {}
logging.info(f"开始比较 {len(model_paths)} 个模型...")
for model_path, model_name in zip(model_paths, model_names):
logging.info(f"评估模型: {model_name} ({model_path})")
try:
# 加载模型
model = load_model(model_path, device=device)
# 评估模型
results = evaluate_model(model, data_loader, device)
# 提取关键指标
key_metrics = {}
if 'regression' in results:
regression_metrics = results['regression']
if 'overall' in regression_metrics:
for metric, value in regression_metrics['overall'].items():
key_metrics[f'regression_{metric}'] = value
if 'calibration' in results:
calibration_metrics = results['calibration']
for metric, value in calibration_metrics.items():
if isinstance(value, (int, float)):
key_metrics[f'calibration_{metric}'] = value
comparison_results[model_name] = {
'model_path': model_path,
'metrics': key_metrics,
'full_results': results
}
except Exception as e:
logging.error(f"评估模型 {model_name} 时发生错误: {e}")
comparison_results[model_name] = {'error': str(e)}
# 生成比较报告
generate_comparison_report(comparison_results, output_dir)
return comparison_results
def generate_comparison_report(comparison_results: Dict[str, Any], output_dir: str):
"""
生成模型比较报告
Args:
comparison_results: 比较结果
output_dir: 输出目录
"""
output_path = Path(output_dir)
output_path.mkdir(parents=True, exist_ok=True)
# 创建比较表格
comparison_data = []
for model_name, results in comparison_results.items():
if 'error' in results:
continue
row = {'Model': model_name}
row.update(results['metrics'])
comparison_data.append(row)
if comparison_data:
df = pd.DataFrame(comparison_data)
# 保存为CSV
df.to_csv(output_path / 'model_comparison.csv', index=False)
# 生成比较图表
if len(comparison_data) > 1:
# 选择关键指标进行比较
key_metrics = ['regression_mae', 'regression_rmse', 'regression_r2', 'calibration_ece']
available_metrics = [m for m in key_metrics if m in df.columns]
if available_metrics:
fig, axes = plt.subplots(2, 2, figsize=(15, 10))
axes = axes.flatten()
for i, metric in enumerate(available_metrics):
if i < len(axes):
ax = axes[i]
# 排序数据
sorted_df = df.sort_values(metric, ascending=metric in ['regression_mae', 'regression_rmse', 'calibration_ece'])
bars = ax.bar(range(len(sorted_df)), sorted_df[metric])
ax.set_xticks(range(len(sorted_df)))
ax.set_xticklabels(sorted_df['Model'], rotation=45, ha='right')
ax.set_ylabel(metric.replace('_', ' ').title())
ax.set_title(f'{metric.replace("_", " ").title()} Comparison')
ax.grid(True, alpha=0.3)
# 添加数值标签
for j, bar in enumerate(bars):
height = bar.get_height()
ax.text(bar.get_x() + bar.get_width()/2., height,
f'{height:.4f}', ha='center', va='bottom')
plt.tight_layout()
plt.savefig(output_path / 'model_comparison.png', dpi=300, bbox_inches='tight')
plt.close()
# 生成文本报告
report_lines = []
report_lines.append("=" * 60)
report_lines.append("模型比较报告")
report_lines.append("=" * 60)
report_lines.append(f"比较时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
report_lines.append(f"模型数量: {len(comparison_results)}")
report_lines.append("")
for model_name, results in comparison_results.items():
report_lines.append(f"模型: {model_name}")
if 'error' in results:
report_lines.append(f" 错误: {results['error']}")
else:
report_lines.append(f" 路径: {results['model_path']}")
for metric, value in results['metrics'].items():
report_lines.append(f" {metric}: {value:.6f}")
report_lines.append("")
report_text = "\n".join(report_lines)
with open(output_path / 'comparison_report.txt', 'w', encoding='utf-8') as f:
f.write(report_text)
logging.info(f"模型比较报告已生成: {output_path}")
def main():
"""主函数"""
# 解析命令行参数
args = parse_arguments()
# 设置日志级别
logging.basicConfig(
level=logging.DEBUG if args.verbose else logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)
logger.info("开始PAD预测器评估")
try:
# 设置设备
if args.device == 'auto':
if torch.cuda.is_available():
device = torch.device(f'cuda:{args.gpu_id}')
logger.info(f"使用GPU: {torch.cuda.get_device_name(args.gpu_id)}")
else:
device = torch.device('cpu')
logger.info("使用CPU")
else:
device = torch.device(args.device)
# 创建输出目录
output_dir = Path(args.output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
# 加载配置
if args.config:
config = load_config(args.config)
else:
config = {
'data': {
'dataloader': {
'batch_size': args.batch_size or 32,
'num_workers': 0,
'pin_memory': False,
'shuffle': False
}
}
}
# 覆盖批次大小
if args.batch_size:
config['data']['dataloader']['batch_size'] = args.batch_size
# 加载模型配置
model_config = None
if args.model_config and os.path.exists(args.model_config):
model_config = load_config(args.model_config)
# 加载数据
data_loader = load_data_for_evaluation(
config,
args.data_path,
args.synthetic_data,
args.num_samples,
args.batch_size
)
if args.compare_models:
# 比较多个模型
logger.info(f"比较 {len(args.compare_models)} 个模型")
comparison_results = compare_models(
args.compare_models,
args.model_names if args.model_names else [],
data_loader,
device,
str(output_dir)
)
logger.info(f"模型比较完成,结果保存在: {output_dir}")
else:
# 评估单个模型
logger.info(f"评估模型: {args.model_path}")
# 加载模型
model = load_model(args.model_path, model_config, device)
# 评估模型
results = evaluate_model(
model,
data_loader,
device,
args.save_predictions,
str(output_dir)
)
# 生成评估报告
report_path = generate_evaluation_report(
results,
str(output_dir),
args.report_name,
args.detailed_analysis,
args.calibration_analysis,
args.error_analysis,
args.generate_plots
)
# 打印关键指标
if 'regression' in results and 'overall' in results['regression']:
overall_metrics = results['regression']['overall']
logger.info("评估结果:")
logger.info(f" MAE: {overall_metrics.get('mae', 0):.6f}")
logger.info(f" RMSE: {overall_metrics.get('rmse', 0):.6f}")
logger.info(f" R²: {overall_metrics.get('r2', 0):.6f}")
logger.info(f" MAPE: {overall_metrics.get('mape', 0):.6f}")
if 'calibration' in results:
calibration_metrics = results['calibration']
logger.info("校准指标:")
logger.info(f" ECE: {calibration_metrics.get('ece', 0):.6f}")
logger.info(f" Sharpness: {calibration_metrics.get('sharpness', 0):.6f}")
logger.info(f"评估完成,报告保存在: {report_path}")
except Exception as e:
logger.error(f"评估过程中发生错误: {e}")
import traceback
logger.error(traceback.format_exc())
sys.exit(1)
if __name__ == "__main__":
main() |