python学习笔记（35)--单特征的时间序列转换为监督学习

通过利用前一时间点的值来预测下一时间点的值，将这个时间序列数据集重构为一个监督学习问题，记一下代码

import numpy as np



# #2.  单特征样本,取一个步长
# dataset = np.array([i for i in range(22)])
# n_dim = dataset.shape[0]
#

# train_size = int(len(dataset) * 7/9)  # 9分之7的数据训练
# validation_size = len(dataset) - train_size  # 9分之2测试数据
# train, validation = dataset[0:train_size], dataset[train_size:len(dataset)]  # 数据切割
#
#
#
# def split_sequence(data,sliding_window_width):
#     X,y = [],[]
#     for i in range(len(data)):
#         end_element_index = i + sliding_window_width
#         if end_element_index > len(data) - 1 :
#             break
#         #这里不用减1,因为列表读数不包含最后一个
#         # x_sequence = data[i:i+sliding_window_width - 1]
#         x_sequence = data[i:end_element_index]
#         y_sequence = data[end_element_index]
#         X.append(x_sequence)
#         y.append(y_sequence)
#     return np.array(X), np.array(y)
#
#
# sw_width = 3
# x_train, y_train = split_sequence(train,sw_width)    #测试
# x_valition, y_valition = split_sequence(validation,sw_width) #验证
#
# print(train_size)
# print(x_train.shape)
# for i in range(len(x_train)):
#     print('X:',x_train[i],'y:',y_train[i])


#3.单特征多步长截取
dataset = [i for i in range (25)]

sw_width = 4
def split_sequnece(sequence,sliding_window_width):
    X,y = [],[]
    for i in range(len(sequence)):
        end_element_index = i + sliding_window_width
        if (end_element_index + 2) > len(sequence) - 1  :
            break

        x_sequence = sequence[i:end_element_index]
        y_sequence = sequence[end_element_index:end_element_index + 2]
        X.append(x_sequence)
        y.append(y_sequence)
    return np.array(X),np.array(y)

#数据集划分，三分之二为训练，三分之一为测试
train_size = int(len(dataset) * 0.67)
validation_size = len(dataset) - train_size

train,validation = dataset[0:train_size],dataset[train_size:len(dataset)]
X_train,y_train = split_sequnece(train,sw_width)

for i in range (len(X_train)):
    print('x:',X_train[i],'y:',y_train[i])