pytorch练习：泰坦尼克号生存预测

学习记录

首先将test.csv,file.csv放同一个文件夹

模型建立

import torch
class Model(torch.nn.Module):
        def __init__(self):
             super(Model, self).__init__()
             #输入的模型特征是6维
             self.linear1 = torch.nn.Linear(6,4)
             self.linear2 = torch.nn.Linear(4,1)
             self.sigmoid = torch.nn.Sigmoid()

        def  forward(self,x):
          
            x = self.sigmoid(self.linear1(x))
            x = self.sigmoid(self.linear2(x))
            return x

加载数据集

import os
os.environ['KMP_DUPLICATE_LIB_OK']='True'
import torch
from torch.utils.data import Dataset,DataLoader

class  TitannikDataset(Dataset):   #需要特征和标签
    def __init__(self, X, y):
        self.len = X.shape[0]
        X = X.values.astype("float32")#.values dataframe to numpy
        y = y.values.astype("float32")
        #    TypeError: expected   np.ndarray(got DataFrame)
        self.x_data  = torch.from_numpy(X)   #numpy to tensor
        self.y_data = torch.from_numpy(y)
    def __getitem__(self,index):
        return self.x_data[index],self.y_data[index]
    def __len__(self):
        return self.len
#加载数据

def  create_train_loader(x_train, y_train):
       dataset = TitannikDataset(x_train, y_train)
       train_loader = DataLoader(dataset=dataset, shuffle=True, batch_size=32, drop_last=True, num_workers=2)
       return train_loader

训练数据并保存模型

import os
os.environ['KMP_DUPLICATE_LIB_OK']='True'
import torch
import ModelDemo
import pandas as pd
from torch.utils.data import Dataset,DataLoader

train = pd.read_csv("train.csv")#此时是dataframe
#取五个最相关的特征
x_train =train[["Pclass", "Sex", "SibSp", "Parch", "Fare"]]
#独热编码
x_train = pd.get_dummies(x_train)#DataFrame
#print(type(x_train))
y_train =train[["Survived"]]#DataFrame
#print(type(y_train))
class  TitannikDataset(Dataset):   #需要特征和标签
    def __init__(self, X, y):
        self.len = X.shape[0]
        X = X.values.astype("float32")#.values dataframe to numpy
        y = y.values.astype("float32")
        #    TypeError: expected   np.ndarray(got DataFrame)
        self.x_data  = torch.from_numpy(X)   #numpy to tensor
        self.y_data = torch.from_numpy(y)
    def __getitem__(self,index):
        return self.x_data[index],self.y_data[index]
    def __len__(self):
        return self.len

#加载数据
model =ModelDemo.Model()
criterion = torch.nn.BCELoss(reduction="mean")
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)
dataset = TitannikDataset(x_train,y_train)#要传入特征 和 标签
train_loader = DataLoader(dataset = dataset,shuffle=True,batch_size=32,drop_last=True,num_workers=2)



def train(epoch):
    #取出小批量
    for i,data in enumerate(train_loader,0):
        input,label =data
        y_pred = model(input)
        loss =   criterion(y_pred,label)

        optimizer.zero_grad()#梯度置零
        loss.backward()
        optimizer.step()
        print("epoch :", epoch, i, loss.item())
        if epoch%200== 199:
            torch.save(model.state_dict(), "model/model_taitan.pth")
            torch.save(optimizer.state_dict(), "model/optimizer_taitan.pth")
            print("epoch :{}训练次数为：{}，损失值为：{}".format(epoch,i, loss.item()))

if __name__ == '__main__':
    for epoch in range(200):
        print({"————————第{}轮测试开始——————".format(epoch+ 1)})
        train(epoch)

进行数据测试

import os
os.environ['KMP_DUPLICATE_LIB_OK']='True'
import torch
import numpy as np
import pandas as pd
from ModelDemo import Model



test = pd.read_csv("test.csv")
x_test =test[["Pclass", "Sex", "SibSp", "Parch", "Fare"]]
x_test = pd.get_dummies(x_test)
x_test_id = test[["PassengerId"]]

model =Model()
if os.path.exists('model/model_taitan.pth'):
    print("ok")
    model.load_state_dict(torch.load("model/model_taitan.pth"))
    #optimizer.load_state_dict(torch.load("model/optimize_taitan.pth"))



def test ():
    with torch.no_grad():
        x_test_ =x_test.values.astype(np.float32)
        x_test_ =torch.Tensor(x_test_)
        #datafrom to numpy   .values
        #numpy to tensor    torch.from_numpy
        y_pred = model (x_test_)
        y_pred_label = torch.where(y_pred>=0.5,torch.tensor([1.0]),torch.Tensor([0.0]))
        y_pred_label = y_pred_label.flatten()#torch.Size([418])
        print(y_pred_label.shape)#torch.Size([418, 1])
        y_pred_label = y_pred_label.to(torch.int64) #提交测试集需要整数
        output = pd.DataFrame({'PassengerId': x_test_id.PassengerId, 'Survived': y_pred_label})
        output.to_csv('my_predict.csv', index=False)

if __name__ == '__main__':
    test()

提交结果