这里好像有点问题

[yueg]

stock-rnn/data_model.py

Line 48 in 2019be9

curr / seq[i][-1] - 1.0 for i, curr in enumerate(seq[1:])]

是不是应该改为
seq = [seq[0] / seq[0][0] - 1.0] + [curr / seq[i-1][-1] - 1.0 for i, curr in enumerate(seq[1:])]

[lilianweng]

Let's check!

from __future__ import division
import numpy as np

def split_and_normalize_prices(seq, input_size):
    seq = [np.array(seq[i * input_size: (i + 1) * input_size]) for i in range(len(seq) // input_size)]
    print "After split:", seq
    seq = [seq[0] / seq[0][0] - 1.0] + [curr / seq[i][-1] - 1.0 for i, curr in enumerate(seq[1:])]
    print "After normalization:", seq
    return seq

With this function defined, split_and_normalize_prices(range(1, 11), 4) prints out:

After split: [array([1, 2, 3, 4]), array([5, 6, 7, 8])]
After normalization: [array([ 0.,  1.,  2.,  3.]), array([ 0.25,  0.5 ,  0.75,  1.  ])]

I believe this is expected behavior :)