<p><strong><em>Split</em></strong>方法在<code>TimeSeriesSplit</code><strong>中生成索引的拆分</strong>。<em>为了得到特定的分割,你需要迭代到它</em>。<em>它被用来迭代时间序列交叉验证的所有可能的分割。在</p>
<p>如果cv分割中测试数据的大小等于<code>s</code>。然后,不管您进行多少次拆分,最后一次拆分集<code>train_data</code>=<code>all data except last s data point</code>和{<cd5>}。所以,如果您想直接进行最后一次拆分:将数据切片。例如,如果数据是一个numpy数组<code>X</code>:</p>
<pre><code>import numpy as np
from sklearn.model_selection import TimeSeriesSplit
X = np.array([[1, 2], [0, 4], [1, 2], [2, 4] ,[1, 2], [7, 4], [8, 2], [5, 4]])
n_splits = 2 # select no of splits required
tscv = TimeSeriesSplit(n_splits = n_splits)
n_samples = X.shape[0] # this is how test_size (s)
s = n_samples//(n_splits + 1) # is evaluated internally
X_train_last, X_test_last = X[ :-s], X[-s: ] # s=2 for this split
X_train_last
# array([[1, 2],
# [0, 4],
# [1, 2],
# [2, 4],
# [1, 2],
# [7, 4]])
X_test_last
# array([[8, 2],
# [5, 4]])
</code></pre>
<p><em>此外,如果您在拆分时设置了“最大列车大小”。那你在切片的时候也要注意这个问题。有关详细信息,请参阅TimeSeriesSplit<a href="http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.TimeSeriesSplit.html" rel="nofollow noreferrer">documentation here</a>。</em></p>