Quick question about LSTM's 3D input shape

Question

I'm very confused about LSTM's 3D input shape.

According to internet, it says that the input shape is [batchsize, timestep, feature].

So if I have a 2D data of 1000 timesteps * 10 features and I would like to predict its future.

Also, if I would like LSTM layer to read 10 timesteps and then predict the next timestep, that is, read t= 1-10, predict t=11.

Does that make my input shape [990,10,10] or [10,990,10]?

Thanks in advance.

Answer 1

The first one is correct. Here's an example:

import tensorflow as tf
import numpy as np

x = np.random.rand(1000, 10)
y = np.random.rand(1000)


def multivariate_data(dataset, target, 
                      start_index, end_index, 
                      history_size, target_size, 
                      step, single_step=False):
  data = []
  labels = []

  start_index = start_index + history_size
  if end_index is None:
    end_index = len(dataset) - target_size

  for i in range(start_index, end_index):
    indices = range(i-history_size, i, step)
    data.append(dataset[indices])

    if single_step:
      labels.append(target[i+target_size])
    else:
      labels.append(target[i:i+target_size])

  return np.array(data), np.array(labels)

x, y = multivariate_data(dataset=x,
                         target=y,
                         start_index=0,
                         end_index=len(x),
                         history_size=10,
                         target_size=1,
                         step=1, single_step=False)

print(x.shape)

(990, 10, 10)