Fix M4 dataset sequence length mismatch

data_provider/data_loader.py

@@ -361,11 +361,22 @@ def __read_data__(self):
             dataset = M4Dataset.load(training=True, dataset_file=self.root_path)
         else:
             dataset = M4Dataset.load(training=False, dataset_file=self.root_path)
-        training_values = np.array(
-            [v[~np.isnan(v)] for v in
-             dataset.values[dataset.groups == self.seasonal_patterns]])  # split different frequencies
-        self.ids = np.array([i for i in dataset.ids[dataset.groups == self.seasonal_patterns]])
-        self.timeseries = [ts for ts in training_values]
+        # training_values = np.array(
+        #     [v[~np.isnan(v)] for v in
+        #      dataset.values[dataset.groups == self.seasonal_patterns]])  # split different frequencies
+        # self.ids = np.array([i for i in dataset.ids[dataset.groups == self.seasonal_patterns]])
+        # self.timeseries = [ts for ts in training_values]
+        mask = dataset.groups == self.seasonal_patterns
+        raw_sequences = [v[~np.isnan(v)] for v in dataset.values[mask]]  # 移除NaN值
+        self.ids = np.array([i for i in dataset.ids[mask]])
+        # 确定目标长度（训练集用seq_len，测试集用M4定义的预测长度）
+        from data_provider.m4 import M4Meta, pad_sequences  # 导入必要的类和函数
+        if self.flag == 'train':
+            target_len = self.seq_len  # 训练集使用模型输入序列长度
+        else:
+            target_len = M4Meta.horizons_map[self.seasonal_patterns]  # 测试集使用M4标准长度
+        # 统一序列长度（填充/截断）
+        self.timeseries = pad_sequences(raw_sequences, target_len, mode='edge')
 
     def __getitem__(self, index):
         insample = np.zeros((self.seq_len, 1))

data_provider/m4.py

@@ -159,3 +159,30 @@ def load_m4_info() -> pd.DataFrame:
     :return: Pandas DataFrame of M4Info.
     """
     # return pd.read_csv(INFO_FILE_PATH)
+
+
+def pad_sequences(sequences, target_len, mode='edge'):
+    """
+    统一序列长度（填充或截断）
+    :param sequences: 原始序列列表（长度可能不同）
+    :param target_len: 目标长度
+    :param mode: 填充模式，'edge' 用最后一个值填充，'zero' 用0填充
+    :return: 长度统一的序列数组
+    """
+    padded = []
+    for seq in sequences:
+        seq_len = len(seq)
+        if seq_len < target_len:
+            # 填充到目标长度
+            pad_length = target_len - seq_len
+            if mode == 'edge':
+                # 用最后一个值填充（更适合时间序列）
+                padded_seq = np.pad(seq, (0, pad_length), mode='edge')
+            else:
+                # 用0填充
+                padded_seq = np.pad(seq, (0, pad_length), mode='constant', constant_values=0)
+        else:
+            # 截断到目标长度
+            padded_seq = seq[:target_len]
+        padded.append(padded_seq)
+    return np.array(padded)