python 株価予想　エラー　Numpy、Pandas、Matplotlib、Pytorch

参考動画
こちらの最後の項目にて、　GAFAの未来の株価予想コードにてエラーが発生してしまい、ご指摘していただきたいのですが。以下のコードが主な一連の流れになります。
ご教授お願いいたいします、

**問題のコード
**```python
fig = plt.figure(figsize=(12, 4))
plt.title('Stock Price Prediction')
plt.ylabel('Price')
plt.grid(True)
plt.autoscale(axis='x', tight=True)
fig.autofmt_xdate()

plt.plot(stock_data_apple["Adj Close"]['2018-01':]*30, label="Apple")
plt.plot(stock_data_amazon["Adj Close"]['2018-01':], label="Amazon")
plt.plot(stock_data_google["Adj Close"]['2018-01':], label="Google")
plt.plot(stock_data_facebook["Adj Close"]['2018-01':]*10, label="Facebook")

plt.legend(bbox_to_anchor=(1.05, 1), loc="upper left", borderaxespad=0, fontsize=18)

plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_apple30)
plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_amazon)
plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_google)
plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_facebook10)

plt.show()

```error
---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-451-c65d3af67862> in <module>()
     25 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_apple*30)
     26 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_amazon)
---> 27 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_google)
     28 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_facebook*10)
     29 

NameError: name 'predicted_labels_array_2d_google' is not defined

↓

python
1import torch
2import torch.nn as nn
3import numpy as np
4import pandas as pd 
5import matplotlib.pyplot as plt

↓

python
1stock_data = pd.read_csv(
2    # "/content/drive/MyDrive/^GSPC.csv",
3    # "/content/drive/MyDrive/AAPL.csv",
4    # "/content/drive/MyDrive/AMZN.csv",
5    # "/content/drive/MyDrive/GOOG.csv",
6    "/content/drive/MyDrive/FB.csv",
7    
8    index_col = 0,
9    parse_dates=True
10)
11
12stock_data

↓

python
1stock_data.drop(
2    ["Open", "High", "Low", "Close", "Volume"],
3    axis="columns",
4    inplace=True
5)
6
7# stock_data
8
9# stock_data_nintendo = stock_data
10# stock_data_sony = stock_data
11
12# stock_data_apple = stock_data
13# stock_data_amazon = stock_data
14# stock_data_google = stock_data
15stock_data_facebook = stock_data

↓

python
1stock_data.plot(figsize=(12, 4))

↓

python
1y = stock_data["Adj Close"].values
2y

↓

python
1from sklearn.preprocessing import MinMaxScaler

↓

python
1scaler = MinMaxScaler(feature_range=(-1, 1))
2scaler.fit(y.reshape(-1, 1))
3y =scaler.transform(y.reshape(-1, 1))
4y

↓

python
1y = torch.FloatTensor(y).view(-1)
2y

↓

python
1# test_size = 24
2test_size = 12
3
4train_seq = y[:-test_size]
5test_seq = y[-test_size:]

↓

pyhton
1plt.figure(figsize=(12, 4))
2plt.xlim(-20, len(test_seq)+20)
3plt.grid(True)
4plt.plot(test_seq)

↓

python
1train_window_size = 12

↓

python
1def input_data(seq, ws):
2  out = []
3  L = len(seq)
4
5  for i in range(L-ws):
6    window = seq[i:i+ws]
7    label = seq[i+ws:i+ws+1]
8    out.append((window, label))
9
10  return out

↓

python
1# train_data = input_data(train_seq, train_window_size)
2train_data = input_data(y, train_window_size)

↓

python
1print("The Number of Training Data: ", len(train_data))

↓

python
1class Model(nn.Module):
2
3  def __init__(self, input=1, h=50, output=1):
4
5    super().__init__()
6    self.hidden_size = h
7
8    self.lstm = nn.LSTM(input, h)
9    self.fc = nn.Linear(h, output)
10
11    self.hidden  = (
12        torch.zeros(1, 1, h),
13        torch.zeros(1, 1, h)
14    )
15
16
17  def forward(self, seq):
18
19    out, _=self.lstm(
20        seq.view(len(seq), 1, -1),
21        self.hidden
22    )
23
24    out = self.fc(
25        out.view(len(seq), -1)
26    )
27
28    return out[-1]

↓

python
1torch.manual_seed(123)
2model = Model()
3criterion = nn.MSELoss()
4optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

↓

python
1# epochs = 10
2epochs = 50
3train_losses = []
4test_losses = []

↓

python
1def run_train():
2    model.train()
3
4    for train_window, correct_label in train_data:
5
6        optimizer.zero_grad()
7
8        model.hidden = (
9            torch.zeros(1, 1, model.hidden_size),
10            torch.zeros(1, 1, model.hidden_size) 
11
12        ) 
13
14        train_predicted_label = model.forward(train_window)
15        train_loss = criterion(train_predicted_label, correct_label)
16
17        train_loss.backward()
18        optimizer.step()
19
20        train_losses.append(train_loss)

↓

python
1a = torch.tensor([3])
2a.item()

↓

python
1def run_test():
2    model.eval()
3
4    for i in range(test_size):
5
6       test_window = torch.FloatTensor(extending_seq[-test_size:])
7
8       with torch.no_grad():
9
10          model.hidden = (
11                torch.zeros(1, 1, model.hidden_size),
12                torch.zeros(1, 1, model.hidden_size)        
13
14          ) 
15          
16          test_predicred_label = model.forward(test_window)
17          extending_seq.append(test_predicred_label.item())
18
19    test_loss = criterion(
20        torch.FloatTensor(extending_seq[-test_size:]),
21        y[len(y)-test_size:]
22    )
23
24    test_losses.append(test_loss)

↓

python
1train_seq[-test_size:]

↓

python
1train_seq[-test_size:].tolist()

↓

python
1for epoch in range(epochs):
2
3    print()
4    print(f'Epoch: {epoch+1}')
5
6    run_train()
7
8    # extending_seq = train_seq[-test_size:].tolist()
9    extending_seq = y[-test_size:].tolist()
10
11    run_test()
12
13    plt.figure(figsize=(12, 4))
14    # plt.xlim(-20, len(y)+20)
15    plt.xlim(-20, len(y)+50)
16    
17    plt.grid(True)
18
19    plt.plot(y.numpy())
20    
21    plt.plot(
22        # range(len(y)-test_size, len(y)),
23        range(len(y), len(y)+test_size),
24        extending_seq[-test_size:]
25    )
26
27    plt.show()

↓

python
1plt.plot(train_losses)

↓

python
1plt.plot(test_losses)

↓

python
1predicted_normalized_labels_list = extending_seq[-test_size:]

↓

python
1predicted_normalized_labels_array_1d = np.array(predicted_normalized_labels_list)
2predicted_normalized_labels_array_1d

↓

python
1predicted_normalized_labels_array_2d = predicted_normalized_labels_array_1d.reshape(-1, 1)
2predicted_normalized_labels_array_2d

↓

python
1predicted_labels_array_2d = scaler.inverse_transform(predicted_normalized_labels_array_2d)
2predicted_labels_array_2d
3
4# predicted_labels_array_2d_nintendo = predicted_labels_array_2d
5# predicted_labels_array_2d_sony = predicted_labels_array_2d
6
7# predicted_labels_array_2d_apple = predicted_labels_array_2d
8# predicted_labels_array_2d_amazon = predicted_labels_array_2d
9# predicted_labels_array_2d_google = predicted_labels_array_2d
10predicted_labels_array_2d_facebook = predicted_labels_array_2d

↓

python
1len(predicted_labels_array_2d)

↓

python
1stock_data["Adj Close"][-test_size:]

↓

python
1len(stock_data["Adj Close"][-test_size:])

↓

python
1stock_data.index

↓

python
1# x_2019_10_to_2020_09 = np.arange('2019-10', '2020-10', dtype='datetime64[M]')
2# x_2019_10_to_2020_09
3
4# x_2019_10_to_2020_09 = np.arange('2019-10', '2020-10', dtype='datetime64[M]')
5# x_2020_10_to_2022_09
6
7# x_2020_10_to_2021_09 = np.arange('2020-10', '2021-10', dtype='datetime64[M]')
8# x_2020_10_to_2021_09
9
10x_2020_11_to_2021_01_weekly = np.arange('2020-11-10', '2021-02-01', dtype='datetime64[W]')
11x_2020_11_to_2021_01_weekly

↓

python
1# len(x_2018_10_to_2020_09)
2# len(x_2019_10_to_2020_09)
3# len(x_2020_10_to_2021_09)
4 
5len(x_2020_11_to_2021_01_weekly)

↓

python
1fig = plt.figure(figsize=(12, 4))
2plt.title('Stock Price Prediction')
3plt.ylabel('Price')
4plt.grid(True)
5plt.autoscale(axis='x', tight=True)
6fig.autofmt_xdate()
7
8plt.plot(stock_data_apple["Adj Close"]['2018-01':]*30, label="Apple")
9plt.plot(stock_data_amazon["Adj Close"]['2018-01':], label="Amazon")
10plt.plot(stock_data_google["Adj Close"]['2018-01':], label="Google")
11plt.plot(stock_data_facebook["Adj Close"]['2018-01':]*10, label="Facebook")
12 
13
14plt.legend(bbox_to_anchor=(1.05, 1), loc="upper left", borderaxespad=0, fontsize=18)
15
16plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_apple*30)
17plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_amazon)
18plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_google)
19plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_facebook*10)
20
21plt.show()

↓

python
1---------------------------------------------------------------------------
2NameError                                 Traceback (most recent call last)
3<ipython-input-451-c65d3af67862> in <module>()
4     25 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_apple*30)
5     26 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_amazon)
6---> 27 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_google)
7     28 plt.plot(x_2020_11_to_2021_01_weekly, predicted_labels_array_2d_facebook*10)
8     29 
9
10NameError: name 'predicted_labels_array_2d_google' is not defined

行動規範の内容に同意します

回答1件

ベストアンサー

単純に、下記のところでpredicted_labels_array_2d_googleがコメントアウトされてるからじゃないでしょうか。

predicted_labels_array_2d = scaler.inverse_transform(predicted_normalized_labels_array_2d)
predicted_labels_array_2d

# predicted_labels_array_2d_nintendo = predicted_labels_array_2d
# predicted_labels_array_2d_sony = predicted_labels_array_2d

# predicted_labels_array_2d_apple = predicted_labels_array_2d
# predicted_labels_array_2d_amazon = predicted_labels_array_2d
# predicted_labels_array_2d_google = predicted_labels_array_2d     #<-----この行
predicted_labels_array_2d_facebook = predicted_labels_array_2d

投稿2020/12/08 11:47

編集2020/12/08 11:48