使用 inverse_transform 返回值错误时出错

Question

我是 LSTM 新手，我正在学习股票价格预测教程。

我正在尝试执行所有操作，但一直收到错误消息：

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-21-87c8b43ab572> in <module>
     36 #Get the models predicted price values
     37 predicted_prices= model.predict(x_test)
---> 38 predicted_prices=scaler.inverse_transform(predicted_prices)
 
~\anaconda3\lib\site-packages\sklearn\preprocessing\_data.py in inverse_transform(self, X)
    430                         force_all_finite="allow-nan")
    431
--> 432         X -= self.min_
    433         X /= self.scale_
    434         return X
 
ValueError: non-broadcastable output operand with shape (401,1) doesn't match the broadcast shape (401,6)

我正在寻找这个错误的解释，这里是代码

company = "AC.TO"
start= dt.datetime(2012,1,1)
end= dt.datetime(2020,1,1)
data = web.DataReader(company, "yahoo", start,end)
data_df= data.filter(['Close'])
dataset= data.values
training_data_len= math.ceil(len(dataset)*.8)
scaler = MinMaxScaler(feature_range=(0,1))
scaled_data = scaler.fit_transform(dataset)
train_data=scaled_data[0:training_data_len, :]
 
predication_days=60
x_train=[]
y_train=[]
 
for x in range (predication_days,len(train_data)):
    x_train.append(train_data[x-predication_days:x, 0])
    y_train.append(train_data[x, 0])
 
    if x<= predication_days+1:
        print (x_train)
        print (y_train)
        print()
x_train, y_train = np.array(x_train), np.array(y_train)
x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1],1))
model= Sequential()
model.add(LSTM(units=50, return_sequences=True, input_shape=(x_train.shape[1],1)))
model.add(LSTM(units=50,return_sequences=False))
model.add(Dense(units=25))
model.add(Dense(units=1))
model.compile(optimizer='adam', loss='mean_squared_error')
model.fit(x_train, y_train, epochs=1, batch_size=1)
test_data = scaled_data[training_data_len - predication_days:,:]
x_test=[]
y_test=dataset[training_data_len:, :]
 
for x in range(predication_days,len(test_data)):
    x_test.append(test_data[x-predication_days:x,0])
 
x_test=np.array(x_test)
x_test=np.reshape(x_test,(x_test.shape[0],x_test.shape[1],1))
predicted_prices= model.predict(x_test)
predicted_prices=scaler.inverse_transform(predicted_prices)

我的问题更多是关于这个错误的原因，从中吸取教训以及如何解决它？

Answer 1

您正在缩放所有数据集列。

因此，每当您进行 inverse_transform 时，都应包含您在 fit_transform 上使用的所有列。

要仅反转标签，您必须分别为您的特征 (X) 和标签 (y) 创建一个比例，然后您可以仅在 y 或 X 中进行 inverse_transform。

另外，在训练数据集上创建scale.fit_transformONLY，然后在测试数据集上创建scale.transform，这是一个很好的做法。

features_scaler = MinMaxScaler(feature_range=(0,1))
label_scaler =  MinMaxScaler(feature_range=(0,1))

train_data = dataset[0:training_data_len, :]
test_data = dataset[training_data_len:, :]
x_train = features_scaler.fit_transform(train_data[my_feature_columns].to_numpy())
y_train = label_scaler.fit_transform(train_data[my_label_column].to_numpy())
x_test = features_scaler.transform(test_data[my_feature_columns].to_numpy())
y_test = label_scaler.transform(test_data[my_label_column].to_numpy())