![](/img/trans.png)
[英]How to train LSTM model with variable sequence lengths and multiple feature dimensions?
[英]How to train a LSTM model with multiple separate training datas?
我有 100 人在 1 年內銷售的銷售數據。
我想讓 ONE 模型預測所有 100 件男裝的銷售情況。
這是我的代碼:
model=Sequential()
y_train=sells_men_sell[1] # sells_men_sell[1] is a 1d array that contains the first sells man's sells record
x_train=sells_men_data[1] # sells_men_sell[1] is a array that contains the first sells man's sells record for training
#, each value in the array(sells_men_sell) contains the sells record for the past 30 days.
model.add(LSTM(50, input_shape=(x_train.shape[1], x_train.shape[2])))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam')
model.fit(x_train, y_train, batch_size=1, epoch=1)
我知道在一個模型中預測 100 個銷售人員聽起來很奇怪,但我正在為一個項目這樣做。
我應該如何處理我的代碼?
我應該在model.fit(x_train, y_train, batch_size=1, epoch=1)
之后添加以下代碼嗎?
y_train1=sells_men_sell[2] # sells_men_sell[2] is a 1d array that contains the second sells man's sells record
x_train1=sells_men_data[2] # sells_men_sell[2] is a array that contains the second sells man's sells record for training
model.add(LSTM(50, input_shape=(x_train1.shape[1], x_train1.shape[2])))
model.fit(x_train1, y_train1, batch_size=1, epoch=1)
您的模型可以有多個輸入以及多個輸出。 您可以使用函數式 API 來實現這一點。
我分享了一個關於如何實現這一目標的小例子。 您可以根據您的用例調整示例。
代碼:
# imports
import tensorflow as tf
import pandas as pd
import numpy as np
# genration of dummy data
x1 = np.random.randint(100, size =(5, 5, 5), dtype = np.int16)
x2 = np.random.randint(100, size =(5, 4, 4), dtype = np.int16)
y1 = np.random.randint(2, size =(5,), dtype = np.int16)
y2 = np.random.randint(2, size =(5,), dtype = np.int16)
# creation of model
def create_model3():
input1 = tf.keras.Input(shape=(5,5,), name = 'I1')
input2 = tf.keras.Input(shape=(4,4,), name = 'I2')
hidden1 = tf.keras.layers.LSTM(units = 4)(input1)
hidden2 = tf.keras.layers.LSTM(units = 4)(input2)
merge = tf.keras.layers.concatenate([hidden1, hidden2])
hidden3 = tf.keras.layers.Dense(units = 3, activation='relu')(merge)
output1 = tf.keras.layers.Dense(units = 2, activation='softmax', name ='O1')(hidden3)
output2 = tf.keras.layers.Dense(units = 2, activation='softmax', name = 'O2')(hidden3)
model = tf.keras.models.Model(inputs = [input1,input2], outputs = [output1,output2])
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
return model
model = create_model3()
tf.keras.utils.plot_model(model, 'my_first_model.png', show_shapes=True)
# training the model
history = model.fit(
x = {'I1':x1, 'I2':x2},
y = {'O1':y1, 'O2': y2},
batch_size = 32,
epochs = 10,
verbose = 1,
callbacks = None,
# validation_data = [(val_data,new_val_data),(val_labels, new_val_labels)]
)
生成的模型如下所示。
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.