活动地址：CSDN21天学习挑战赛
参考文章：https://mtyjkh./article/details/117752046
一、RNN（循环神经网络）介绍
传统的神经网络的结构比较简单：输入层——隐藏层——输出层
RNN跟传统神经网络最大的区别在于，每次都会将前一次的输出结果带到下一次的隐藏层中，一起训练。如下图所示：
这里用一个具体的案例来看看RNN是如何工作的：
用户说了一句“what time is it？”，我们的神经网络会将这句话分成五个基本单元（十个单词+一个问号）
然后，按照顺序将五个基本单元输入RNN网络，先将“what”作为RNN的输入，得到01
随后，按照顺序将“time”输入到RNN网络，得到02。
这个过程我们可以看到，输入“time”的时候，前面“what”的输出也会对02的输出产生了影响（隐藏层中有一半是黑色的）。
以此类推，我们可以看到，前面所有的输入产生的结果都对后续的输出产生了影响（可以看到圆形中包含了前面所有的颜色）
当神经网络判断亿图的时候，只需要最后一层的输出05即可，如下图所示：
二、准备工作
1. 设置GPU
如果使用的是CPU，可以不设置此部分。
import tensorflow as tfgpus = tf.config.list_physical_devices("GPU")if gpus:tf.config.experimental.set_memory_growth(gpus[0], True) # 设置GPU显存按需使用tf.config.set_visible_devices([gpus[0]], "GPU")
2. 加载数据
import os, mathfrom tensorflow.keras.layers import Dropout, Dense, SimpleRNNfrom sklearn.preprocessing import MinMaxScalerfrom sklearn import metricsimport numpy as npimport pandas as pdimport tensorflow as tfimport matplotlib.pyplot as plt# 设置图表的显示支持中文plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号
data = pd.read_csv('./datasets/SH600519.csv') # 读取股票文件data
# 前（2426-300=2126）天的开盘价作为训练集，表格从0开始计数# 2:3是提取[2:3)列，左闭右开# 最后300天的开盘价作为测试集training_set =data.iloc[0: 2426 - 300, 2: 3].valuestest_set = data.iloc[2426-300: , 2: 3].values
三、数据预处理
1. 归一化
sc = MinMaxScaler(feature_range=(0, 1))training_set = sc.fit_transform(training_set)test_set = sc.transform(test_set)
2. 设置测试集和训练集
x_train = []y_train = []x_test = []y_test = []# 使用前60天的开盘价作为输入特征x_train# 第61天的开盘价作为输入标签y_train# for循环共构建2426-300-60=2066组训练数据# 共构建300-60=260组测试数据for i in range(60, len(training_set)):x_train.append(training_set[i - 60 : i, 0])y_train.append(training_set[i, 0])for i in range(60, len(test_set)):x_test.append(test_set[i - 60 : i, 0])y_test.append(test_set[i, 0])# 对训练集进行打乱np.random.seed(7)np.random.shuffle(x_train)np.random.seed(7)np.random.shuffle(y_train)tf.random.set_seed(7)
# 将训练数据调整为数组# 调整后的形状：# x_train:(2066, 60, 1)# y_train:(2066, )# x_test:(240, 60, 1)# y_test:(240, )x_train, y_train = np.array(x_train), np.array(y_train) x_test, y_test = np.array(x_test), np.array(y_test)# 输入要求：[送入样本数，循环核时间展开步数，每个时间步输入特征个数]x_train = np.reshape(x_train, (x_train.shape[0], 60, 1))x_test = np.reshape(x_test, (x_test.shape[0], 60, 1))
五、构建模型
model = tf.keras.Sequential([SimpleRNN(100, return_sequences=True), # 布尔值。判断是返回输出序列中的最后一个输出，还是全部序列Dropout(0.1), # 防止过拟合SimpleRNN(100),Dense(1)])
六、激活模型
# 该应用只观测loss数值，不观测准确率，所以删去metrics选项，后面在每个epoch迭代显示时只显示loss值pile(optimizer=tf.keras.optimizers.Adam(0.001),loss='mean_squared_error') # 损失函数用均方误差
七、训练模型
history = model.fit(x_train, y_train,batch_size=64,epochs=35,validation_data=(x_test, y_test),validation_freq=1) # 测试的epoch间隔数model.summary()
Epoch 1/3533/33 [==============================] - 1s 21ms/step - loss: 2.1443e-04 - val_loss: 0.0191Epoch 2/3533/33 [==============================] - 1s 20ms/step - loss: 1.7079e-04 - val_loss: 0.0180Epoch 3/3533/33 [==============================] - 1s 20ms/step - loss: 1.8806e-04 - val_loss: 0.0270Epoch 4/3533/33 [==============================] - 1s 21ms/step - loss: 1.8641e-04 - val_loss: 0.0212Epoch 5/3533/33 [==============================] - 1s 20ms/step - loss: 1.7237e-04 - val_loss: 0.0220Epoch 6/3533/33 [==============================] - 1s 20ms/step - loss: 1.9482e-04 - val_loss: 0.0214Epoch 7/3533/33 [==============================] - 1s 21ms/step - loss: 2.2625e-04 - val_loss: 0.0269Epoch 8/3533/33 [==============================] - 1s 21ms/step - loss: 1.8843e-04 - val_loss: 0.0318Epoch 9/3533/33 [==============================] - 1s 21ms/step - loss: 2.9509e-04 - val_loss: 0.0231Epoch 10/3533/33 [==============================] - 1s 21ms/step - loss: 2.5584e-04 - val_loss: 0.0126Epoch 11/3533/33 [==============================] - 1s 21ms/step - loss: 1.6293e-04 - val_loss: 0.0141Epoch 12/3533/33 [==============================] - 1s 21ms/step - loss: 1.8390e-04 - val_loss: 0.0147Epoch 13/3533/33 [==============================] - 1s 21ms/step - loss: 1.7752e-04 - val_loss: 0.0186Epoch 14/3533/33 [==============================] - 1s 21ms/step - loss: 2.1432e-04 - val_loss: 0.0205Epoch 15/3533/33 [==============================] - 1s 21ms/step - loss: 2.1611e-04 - val_loss: 0.0093Epoch 16/3533/33 [==============================] - 1s 20ms/step - loss: 2.0771e-04 - val_loss: 0.0245Epoch 17/3533/33 [==============================] - 1s 21ms/step - loss: 2.5106e-04 - val_loss: 0.0106Epoch 18/3533/33 [==============================] - 1s 21ms/step - loss: 1.9776e-04 - val_loss: 0.0173Epoch 19/3533/33 [==============================] - 1s 21ms/step - loss: 1.7719e-04 - val_loss: 0.0247Epoch 20/3533/33 [==============================] - 1s 21ms/step - loss: 2.1179e-04 - val_loss: 0.0298Epoch 21/3533/33 [==============================] - 1s 21ms/step - loss: 1.9824e-04 - val_loss: 0.0147Epoch 22/3533/33 [==============================] - 1s 21ms/step - loss: 2.0879e-04 - val_loss: 0.0260Epoch 23/3533/33 [==============================] - 1s 21ms/step - loss: 1.7415e-04 - val_loss: 0.0176Epoch 24/3533/33 [==============================] - 1s 21ms/step - loss: 1.6353e-04 - val_loss: 0.0090Epoch 25/3533/33 [==============================] - 1s 21ms/step - loss: 2.1351e-04 - val_loss: 0.0076Epoch 26/3533/33 [==============================] - 1s 21ms/step - loss: 1.7860e-04 - val_loss: 0.0170Epoch 27/3533/33 [==============================] - 1s 21ms/step - loss: 1.6161e-04 - val_loss: 0.0175Epoch 28/3533/33 [==============================] - 1s 21ms/step - loss: 1.5730e-04 - val_loss: 0.0108Epoch 29/3533/33 [==============================] - 1s 22ms/step - loss: 1.5606e-04 - val_loss: 0.0141Epoch 30/3533/33 [==============================] - 1s 22ms/step - loss: 1.7033e-04 - val_loss: 0.0119Epoch 31/3533/33 [==============================] - 1s 22ms/step - loss: 1.7409e-04 - val_loss: 0.0164Epoch 32/3533/33 [==============================] - 1s 21ms/step - loss: 1.6120e-04 - val_loss: 0.0168Epoch 33/3533/33 [==============================] - 1s 21ms/step - loss: 1.6100e-04 - val_loss: 0.0238Epoch 34/3533/33 [==============================] - 1s 21ms/step - loss: 1.5991e-04 - val_loss: 0.0299Epoch 35/3533/33 [==============================] - 1s 21ms/step - loss: 1.8989e-04 - val_loss: 0.0176Model: "sequential"_________________________________________________________________Layer (type) Output Shape Param # =================================================================simple_rnn (SimpleRNN) (None, 60, 100) 10200_________________________________________________________________dropout (Dropout) (None, 60, 100) 0 _________________________________________________________________simple_rnn_1 (SimpleRNN)(None, 100)0_________________________________________________________________dense (Dense)(None, 1) 101 =================================================================Total params: 30,401Trainable params: 30,401Non-trainable params: 0
八、结果可视化
1. 绘制loss图
plt.plot(history.history['loss'], label="Training Loss")plt.plot(history.history['val_loss'], label='Validation Loss')plt.title("Training and Validation Loss")plt.legend()plt.show()
2. 预测
predicted_stock_price = model.predict(x_test) # 测试集输入模型进行预测predicted_stock_price = sc.inverse_transform(predicted_stock_price) # 对预测数据还原——从（0,1）反归一化到原始范围real_stock_price = sc.inverse_transform(test_set[60: ]) # 对真实数据还原——从（0,1）反归一化到原始范围# 画出真实数据和预测数据的对比曲线plt.plot(real_stock_price, color='red', label='Stock Price')plt.plot(predicted_stock_price, color='blue', label='Predicted Stock Price')plt.title('Stock Price Prediction')plt.xlabel('Time')plt.ylabel('Stock Price')plt.legend()plt.show()
3. 评估
# MSE：均方误差——预测值减真实值求平方后求均值# RMSE：均方根误差——对均方误差开方# MAE：平均绝对误差——预测值减真实值求绝对值后求均值# R2：决定系数——可简单理解为反映模型拟合优度的重要的统计量# 参考文章：/qq_38251616/article/details/107997435MSE = metrics.mean_squared_error(predicted_stock_price, real_stock_price)RMSE = metrics.mean_squared_error(predicted_stock_price, real_stock_price) ** 0.5MAE = metrics.mean_absolute_error(predicted_stock_price, real_stock_price)R2 = metrics.r2_score(predicted_stock_price, real_stock_price)print("均方误差：%.5f" % MSE)print("均方根误差：%.5f" % RMSE)print("平均绝对误差：%.5f" % MAE)print("R2：%.5f" % R2)
