在以数组形式提供的时间序列上创建滑动窗口的数据集。
导入库
tf.keras.utils.timeseries_dataset_from_array
tf.keras.utils.timeseries_dataset_from_array(
data,
targets,
sequence_length,
sequence_stride=1,
sampling_rate=1,
batch_size=128,
shuffle=False,
seed=None,
start_index=None,
end_index=None
)
生成数据
data = [i for i in range(100)]
np.array(data)
array([ 0, 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])
一对一的关系
生成使用过去 10 个数据集 时间步长 要预测下一个时间步长
input_data = data[:-10]
targets = data[10:]
dataset = keras.preprocessing.timeseries_dataset_from_array(
input_data, targets, sequence_length=10)
for batch in dataset:
# print(batch)
inputs, targets = batch
assert np.array_equal(inputs[0], data[:10]) # First sequence: steps [0-9]
# Corresponding target: step 10
assert np.array_equal(targets[0], data[10])
break
查看输入数据
list(dataset.as_numpy_iterator())[0][0]
另外一种查看数据的方法
inputs.numpy()
多对一的关系
生成数据
data = [i for i in range(100)]
data = np.array(data).reshape(25,4)
data
切分输入和输出数据
input_data = data[:end,:]
input_data
targets_data = data[start:,1]
targets_data
为方便理解。将输入输出数据可视化:
其中0-3为输入变量,4为预测变量
多对多的关系
sample_length = 5
input_dataset = tf.keras.utils.timeseries_dataset_from_array(data[:-sample_length,:], None, sequence_length=5,sequence_stride=sample_length)
target_dataset = tf.keras.utils.timeseries_dataset_from_array(
data[sample_length:,1], None, sequence_length=sample_length, sequence_stride=5)
表格可视化
list(input_dataset.as_numpy_iterator())[0][0]
#%%
list(target_dataset.as_numpy_iterator())[0]