神经网络（数据思维赛-家电能源预测）

训练集给出如下数据：

 测试集提供其中的部分列：

 

要求预测以下列的数据：

['Tdewpoint', 'Visibility', 'Windspeed', 'RH_out', 'Press_mm_hg', 'RH_9', 'T_out', 'RH_4']

 

使用神经网络进行预测：

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from keras.models import Model
from keras.layers import Input, Dense

# 读取训练集数据
train_data = pd.read_csv('train_dataset.csv')

li=train_data.columns.to_list()[2::]
#print(li)
goal=['Tdewpoint', 'Visibility', 'Windspeed', 'RH_out', 'Press_mm_hg', 'RH_9', 'T_out', 'RH_4']
feature=list(set(li)-set(goal))
#print(feature)
#exit()

# 选择特征和目标变量
X = train_data[feature]
y = train_data[goal]

# 划分训练集和验证集
X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

# 构建神经网络模型
input_layer = Input(shape=(len(feature),))
hidden_layer = Dense(10, activation='relu')(input_layer)
output_layer1 = Dense(1, name='output1')(hidden_layer)
output_layer2 = Dense(1, name='output2')(hidden_layer)
output_layer3 = Dense(1, name='output3')(hidden_layer)
output_layer4 = Dense(1, name='output4')(hidden_layer)
output_layer5 = Dense(1, name='output5')(hidden_layer)
output_layer6 = Dense(1, name='output6')(hidden_layer)
output_layer7 = Dense(1, name='output7')(hidden_layer)
output_layer8 = Dense(1, name='output8')(hidden_layer)
model = Model(inputs=input_layer, outputs=[output_layer1, output_layer2, output_layer3, output_layer4, output_layer5, output_layer6, output_layer7, output_layer8])
model.compile(loss='mse', optimizer='adam', metrics=['mse'])

# 训练模型
model.fit(X_train, [y_train[c] for c in y_train.columns], epochs=1000, batch_size=32, verbose=1, validation_data=(X_val, [y_val[c] for c in y_val.columns]))

# 读取测试集数据
test_data = pd.read_csv('test_dataset.csv')

# 预测并保存结果
X_test = test_data[feature]
y_pred = model.predict(X_test)
submission = pd.DataFrame({'Tdewpoint': y_pred[0].flatten(), 'Visibility': y_pred[1].flatten(), 'Windspeed': y_pred[2].flatten(), 'RH_out': y_pred[3].flatten(), 'Press_mm_hg': y_pred[4].flatten(), 'RH_9': y_pred[5].flatten(), 'T_out': y_pred[6].flatten(), 'RH_4': y_pred[7].flatten()})
submission.to_csv('test_result.csv', index=False)