介绍：

交叉验证_百度百科

时序算法模型的交叉验证：

需求：我们在进行机器学习算法，为了保能够得到可靠稳定的模型，通常需要用到交叉验证法来对模型进行验证。常见的交叉验证形式有Holdout验证，K-fold验证，留一验证等。

与其他机器学习算法不同，时序算法由于数据具有前后关系，因此使用KFold方法无法满足要求，本文将介绍一种利用sklearn库中的TimeSeriesSplit方法，以下是以LSTM模型预测股票数据为例子，采用交叉验证法的具体代码实现。

交叉验证法：

import pandas as pdimport numpy as np import matplotlib.pyplot as pltfrom sklearn.preprocessing import MinMaxScalerfrom keras.models import Sequentialfrom keras.layers import LSTM,Dense,Dropout,Bidirectional,Activationfrom keras.optimizers import Adamstock = pd.DataFrame(pd.read_csv('stock2.csv'))# 完整的数据集，# 交叉验证我们需要先根据股票数据根据time_stamp，得到全部的数据集；# 这里的做法和你上面的基本类似scaler = MinMaxScaler(feature_range=(0, 1)) scaled_data = scaler.fit_transform(stock)X, Y = [], []print(scaled_data)# 训练集#print(scaled_data.shape)#print(scaled_data[1, 3])for i in range(time_stamp, len(stock)):X.append(scaled_data[i - time_stamp:i])Y.append(scaled_data[i,3])X, Y = np.array(X), np.array(Y)

# 每一个数据集需要有一个单独的模型去训练,因此定义一个产生lstm模型的函数def build_model(epochs = 50,batch_size = 16):# 超参数# LSTM 参数: return_sequences=True LSTM输出为一个序列。默认为False，输出一个值。# input_dim：输入单个样本特征值的维度# input_length：输入的时间点长度model = Sequential()model.add(Bidirectional(LSTM(units=100, return_sequences=True, input_dim=x_train.shape[-1],input_length=x_train.shape[1])))model.add(Dropout(0.5))model.add(LSTM(units=80))model.add(Dropout(0.5))model.add(Dense(50))#model.add(Activation('tanh'))model.add(Dense(1))# pile(loss='mean_squared_error', optimizer=tf.keras.optimizers.Adam(0.01))pile(loss='mean_squared_error', optimizer=Adam(lr = 0.1))return model

# 导入sklearn库中用于时序模型K折扣交叉验证的TimeSeriesSplitfrom sklearn.model_selection import KFoldfrom sklearn.model_selection import TimeSeriesSplittscv=TimeSeriesSplit(gap=0, max_train_size=None, n_splits=5, test_size=None)rmse_list=[] # 勇于保存每次的测试误差# 遍历,每一次取出对应的训练和测试集的indexfor kfold, (train_index, train_index) in enumerate(tscv.split(X)):print('train_index', train_index, 'test_index', test_index)# 根据索引得到对应的训练集和测试集train_X, train_y = X[train_index], Y[train_index]test_X, test_y = X[test_index], Y[test_index]# 建立模型并训练model=build_model(epochs = 5,batch_size = 128)history = model.fit(train_X, train_y, epochs=epochs, batch_size=batch_size,validation_split=0.1,verbose=2)# 计算测试集误差test_pred = model.predict(test_X)scaler.fit_transform(pd.DataFrame(test['Close'].values))test_pred = scaler.inverse_transform(test_pred)test_y = scaler.inverse_transform([test_y])rmse = np.sqrt(np.mean(np.power((test_y - test_pred), 2)))rmse_list.append(rmse)print('rmse of %d fold=%.4f'%(kfold,rmse))# 总的误差为每次误差的均值print('average rmse:',np.mean(rmse_list))

以下是sklearn官方给的一个例子：

>>> import numpy as np>>> from sklearn.model_selection import TimeSeriesSplit>>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4], [1, 2], [3, 4]])>>> y = np.array([1, 2, 3, 4, 5, 6])>>> tscv = TimeSeriesSplit()>>> print(tscv)TimeSeriesSplit(gap=0, max_train_size=None, n_splits=5, test_size=None)>>> for train_index, test_index in tscv.split(X):...print("TRAIN:", train_index, "TEST:", test_index)...X_train, X_test = X[train_index], X[test_index]...y_train, y_test = y[train_index], y[test_index]TRAIN: [0] TEST: [1]TRAIN: [0 1] TEST: [2]TRAIN: [0 1 2] TEST: [3]TRAIN: [0 1 2 3] TEST: [4]TRAIN: [0 1 2 3 4] TEST: [5]>>> # Fix test_size to 2 with 12 samples>>> X = np.random.randn(12, 2)>>> y = np.random.randint(0, 2, 12)>>> tscv = TimeSeriesSplit(n_splits=3, test_size=2)>>> for train_index, test_index in tscv.split(X):... print("TRAIN:", train_index, "TEST:", test_index)... X_train, X_test = X[train_index], X[test_index]... y_train, y_test = y[train_index], y[test_index]TRAIN: [0 1 2 3 4 5] TEST: [6 7]TRAIN: [0 1 2 3 4 5 6 7] TEST: [8 9]TRAIN: [0 1 2 3 4 5 6 7 8 9] TEST: [10 11]>>> # Add in a 2 period gap>>> tscv = TimeSeriesSplit(n_splits=3, test_size=2, gap=2)>>> for train_index, test_index in tscv.split(X):... print("TRAIN:", train_index, "TEST:", test_index)... X_train, X_test = X[train_index], X[test_index]... y_train, y_test = y[train_index], y[test_index]TRAIN: [0 1 2 3] TEST: [6 7]TRAIN: [0 1 2 3 4 5] TEST: [8 9]TRAIN: [0 1 2 3 4 5 6 7] TEST: [10 11]

如果觉得《时序算法交叉验证法python实现》对你有帮助，请点赞、收藏，并留下你的观点哦！