将时间步长值更改为 Keras LSTM 的变量

Question

问题很简单，我已经在互联网上广泛搜索了解决方案。如何重塑数据框以具有我为 LSTM 提供的自定义时间步长？

look_back=90
train_X = np.reshape(
    train_dataset, (train_dataset.shape[0], look_back, train_dataset.shape[1]))
test_X = np.reshape(
    test_dataset, (test_dataset.shape[0], look_back, `test_dataset.shape[1]))

以上抛出以下错误 ValueError: cannot reshape array of size 446208 into shape (3984,90,112) 那么，如何将回溯更改为变量而不是 1，从而使 LSTM 的内存功能变得无用。

编辑使用下面提供的解决方案，我将代码更改为

## train test split
train_split = 0.8
train_size = int(n_sample*train_split)

X_train = X[:train_size] # (train_size, n_features)
X_test = X[train_size:] # (n_sample-train_size, n_features)
print(X_train.shape, X_test.shape) 
y_train = y[:train_size] # (train_size,)
y_test = y[train_size:] # (n_sample-train_size,)
print(y_train.shape, y_test.shape) 

look_back = 90
y_train = y_train[look_back:] # (train_size-look_back,)
y_test = y_test[look_back:] # ((n_sample-train_size),)
print(y_train.shape, y_test.shape)
X_train = view_as_windows(X_train, (look_back,n_features))[:-1,0] # (train_size-look_back, look_back, n_features)
X_test = view_as_windows(X_test, (look_back,n_features))[:-1,0] # ((n_sample-train_size)-look_back, look_back, n_features)
print(X_train.shape, X_test.shape)

它现在很有魅力！

Answer 1

这是我建议你创建滑动的技巧windows...

import numpy as np
from skimage.util.shape import view_as_windows

## create dummy data
n_sample = 2000
n_features = 5

X = np.tile(np.arange(n_sample), (n_features,1)).T

X_train = X[:int(n_sample*0.8)]
X_test = X[int(n_sample*0.8):]

## create windows
look_back = 90
X_train = view_as_windows(X_train, (look_back,n_features), step=1)[:-1,0]
X_test = view_as_windows(X_test, (look_back,n_features), step=1)[:-1,0]
print(X_train.shape, X_test.shape) # (1510, 90, 5) (310, 90, 5)

将时间步长值更改为 Keras LSTM 的变量

Changing the time step value to a variable for a Keras LSTM

python

machine-learning

lstm

keras

tensorflow