也不知道对不对，就凭着自己的思路写了一个
数据集：https://www.kaggle.com/c/titanic/data
import torch
import torch.nn as nn
import pandas as pd
import numpy as np class DataProcessing(object):
def __init__(self):
pass def get_data(self):
data_train = pd.read_csv('train.csv')
label = data_train[['Survived']]
data_test = pd.read_csv('test.csv')
# 读取指定列
gender = pd.read_csv('gender_submission.csv', usecols=[1])
return data_train, label, data_test, gender def data_processing(self, data_):
# 训练集测试集都进行相同的处理
data = data_[['Pclass', 'Sex', 'Age', 'SibSp', 'Fare', 'Cabin', 'Embarked']]
data['Age'] = data['Age'].fillna(data['Age'].mean())
data['Cabin'] = pd.factorize(data.Cabin)[0]
data.fillna(0, inplace=True)
data['Sex'] = [1 if x == 'male' else 0 for x in data.Sex]
data['p1'] = np.array(data['Pclass'] == 1).astype(np.int32)
data['p2'] = np.array(data['Pclass'] == 2).astype(np.int32)
data['p3'] = np.array(data['Pclass'] == 3).astype(np.int32)
data['e1'] = np.array(data['Embarked'] == 'S').astype(np.int32)
data['e2'] = np.array(data['Embarked'] == 'C').astype(np.int32)
data['e3'] = np.array(data['Embarked'] == 'Q').astype(np.int32)
del data['Pclass']
del data['Embarked']
return data def data(self):
# 读数据
train_data, label, test_data, gender = self.get_data()
# 处理数据
# 训练集输入数据
train = np.array(data_processing.data_processing(train_data))
# 训练集标签
train_label = np.array(label)
# 测试集
test = np.array(data_processing.data_processing(test_data))
# 测试集标签
test_label = np.array(gender) train = torch.from_numpy(train).float()
train_label = torch.tensor(train_label).float()
test = torch.tensor(test).float()
test_label = torch.tensor(test_label) return train, train_label, test, test_label class MyNet(nn.Module):
def __init__(self):
super(MyNet, self).__init__()
self.fc = nn.Sequential(
nn.Linear(11, 7),
nn.Sigmoid(),
nn.Linear(7, 7),
nn.Sigmoid(),
nn.Linear(7, 1),
)
self.opt = torch.optim.Adam(params=self.parameters(), lr=0.001)
self.mls = nn.MSELoss() def forward(self, inputs):
# 前向传播
return self.fc(inputs) def train(self, inputs, y):
# 训练
out = self.forward(inputs)
loss = self.mls(out, y)
self.opt.zero_grad()
loss.backward()
self.opt.step()
# print(loss) def test(self, x, y):
# 测试
# 将variable张量转为numpy
# out = self.fc(x).data.numpy()
count = 0
out = self.fc(x)
sum = len(y)
for i, j in zip(out, y):
i = i.detach().numpy()
j = j.detach().numpy()
loss = abs((i - j)[0])
if loss < 0.3:
count += 1
# 误差0.3内的正确率
print(count/sum) if __name__ == '__main__':
data_processing = DataProcessing()
train_data, train_label, test_data, test_label = data_processing.data()
net = MyNet()
count = 0
for i in range(20000):
# 为了减小电脑压力,分批训练 100个训练一次 ## 2018.12.22补充：正确的做法应该是用batch
for n in range(len(train_data)//100 + 1):
batch_data = train_data[n*100: n*100 + 100]
batch_label = train_label[n*100: n*100 + 100]
net.train(train_data, train_label)
net.test(test_data, test_label) # 输出结果：0.7488038277511961
效果一般吧，不过至少出来了，hiahiahia