Pytorch forecast house price

import numpy as np
import pandas as pd
import torch
from torch import nn
from d2l import torch as d2l
import matplotlib.pyplot as plt

#  Read training set data 
train_data = pd.read_csv('../data/train.csv')
#  Read test set data 
test_data = pd.read_csv('../data/test.csv')
#  Print dataset size 
# print(train_data.shape)
# print(test_data.shape)

#  Data preprocessing . Delete the first row of training set data ID And the target price in the last line , Delete the first row of test set data ID, The results are combined as features 
all_features = pd.concat((train_data.iloc[:, 1:-1], test_data.iloc[:, 1:]))
#print(all_features.dtypes)
#  Get non object Data column index of type 
features_index = all_features.dtypes[all_features.dtypes != 'object'].index
#  Standardized data 
all_features[features_index] = all_features[features_index].apply(lambda x: (x - x.mean()) / x.std())
#  The missing value （NaN） Set to 0
all_features[features_index] = all_features[features_index].fillna(0)
#  Hot coding alone 
all_features = pd.get_dummies(all_features, dummy_na=True)
print(all_features.shape)

#  Get the number of training dataset rows 
train_count = train_data.shape[0]
print(train_count)
#  Get the characteristics of the training data set 
train_features = torch.tensor(all_features[:train_count].values, dtype=torch.float32)
#  Get the characteristics of the test data set 
test_features = torch.tensor(all_features[train_count:].values, dtype=torch.float32)
#  Get the label of the training data set 
train_labels = torch.tensor(train_data.SalePrice.values.reshape(-1, 1), dtype=torch.float32)

#  The loss function is defined as the mean square loss function 
loss_function = nn.MSELoss()

#  Multilayer perceptron model 
class Net(nn.Module):
    def __init__(self, in_features):
        super(Net, self).__init__()
        self.linear_relu1 = nn.Linear(in_features, 128) #  Input layer 
        self.linear_relu2 = nn.Linear(128, 256) #  Hidden layer 
        self.linear_relu3 = nn.Linear(256, 256) #  Hidden layer 
        self.linear_relu4 = nn.Linear(256, 256) #  Hidden layer 
        self.linear5 = nn.Linear(256, 1) #  Output layer 

    def forward(self, x):
        y_pred = self.linear_relu1(x)
        y_pred = nn.functional.relu(y_pred)

        y_pred = self.linear_relu2(y_pred)
        y_pred = nn.functional.relu(y_pred)

        y_pred = self.linear_relu3(y_pred)
        y_pred = nn.functional.relu(y_pred)

        y_pred = self.linear_relu4(y_pred)
        y_pred = nn.functional.relu(y_pred)

        y_pred = self.linear5(y_pred)
        return y_pred

multilayer_perceptrons_model = Net(train_features.shape[1])

#  Loss function ： Logarithmic mean square function 
def log_rmse(net, features, labels):
    #  The output result of the model is less than 1 Is set to 1. The purpose is to further stabilize the result when taking logarithm 
    clipped_preds = torch.clamp(net(features), 1, float('inf'))
    rmse = torch.sqrt(loss_function(torch.log(clipped_preds), torch.log(labels)))
    return rmse.item()

#  model training 
def train(net, train_features, train_labels, test_features, test_labels,
          num_epochs, learning_rate, weight_decay, batch_size):
    train_ls, test_ls = [], []
    train_iter = d2l.load_array((train_features, train_labels), batch_size)
    #  Optimizer usage Adam optimization algorithm 
    optimizer = torch.optim.Adam(net.parameters(), lr = learning_rate, weight_decay = weight_decay)

    #  Model iteration training 
    for epoch in range(num_epochs):
        #  Train a set of parameters ？
        for X, y in train_iter:
            optimizer.zero_grad()
            loss = loss_function(net(X), y)
            loss.backward()
            optimizer.step()
        train_ls.append(log_rmse(net, train_features, train_labels))
        if test_labels is not None:
            test_ls.append(log_rmse(net, test_features, test_labels))
    
    return train_ls, test_ls

#  obtain K Fold cross validation dataset . Put the training set data into the atmosphere k Share , Among them the first i Copies are used as validation data , The rest are used as training data 
def get_k_fold_data(k, i, X, y):
    assert k > 1
    fold_size = X.shape[0] // k #  to be divisible by k
    X_train, y_train = None, None
    for j in range(k):
        idx = slice(j * fold_size, (j + 1) * fold_size)
        X_part, y_part = X[idx, :], y[idx]
        if j == i:
            X_valid, y_valid = X_part, y_part
        elif X_train is None:
            X_train, y_train = X_part, y_part
        else:
            X_train = torch.cat([X_train, X_part], 0)
            y_train = torch.cat([y_train, y_part], 0)
    return X_train, y_train, X_valid, y_valid

#  be based on K Fold cross validation dataset execution K Time training 
def k_fold_train(k, X_train, y_train, num_epochs, learning_rate, weight_decay, batch_size):
    train_l_sum, valid_l_sum = 0, 0
    for i in range(k):
        # data = get_k_fold_data(k, i, X_train, y_train)
        train_features, train_labels, valid_features, valid_labels = get_k_fold_data(k, i, X_train, y_train)
        #  This *data What do you mean ？？？
        # train_ls, valid_ls = train(net, *data, num_epochs, learning_rate, weight_decay, batch_size)
        train_ls, valid_ls = train(multilayer_perceptrons_model, train_features, train_labels, valid_features, valid_labels, num_epochs, learning_rate, weight_decay, batch_size)
        train_l_sum += train_ls[-1]
        valid_l_sum += valid_ls[-1]
        #  Draw a subgraph 
        plt.subplot(2, 3, i + 1)
        train_line, = plt.plot(list(range(1, num_epochs + 1)), train_ls)
        valid_line, = plt.plot(valid_ls)        
        plt.xlabel("epoch")
        plt.ylabel("rmse")
        # plt.yscale('log')
        plt.title('#{} fold result'.format(i + 1))
        plt.legend([train_line, valid_line], ['train', 'valid'], loc='best')
        plt.grid(False)
        print(f' fold {i + 1},  Training log rmse={float(train_ls[-1]):f}, '
              f' verification log rmse={float(valid_ls[-1]):f}')
    return train_l_sum / k, valid_l_sum / k

k = 5
num_epochs = 100
lr = 1e-4
weight_decay = 0
batch_size = 64
#  Training 
train_l, valid_l = k_fold_train(k, train_features, train_labels, num_epochs, lr, weight_decay, batch_size)
print(f'{k}- Fold validation ： Average training log rmse：{float(train_l):f}, '
      f' Average validation log rmse：{float(valid_l):f}')

#  Save model parameters 
torch.save(multilayer_perceptrons_model.state_dict(), 'model_param')

#  Show drawing 
plt.tight_layout()
plt.show()

• Environmental Science

OS: macOS 12.1

pytorch: 1.10.1

• C++ call pytorch Model

• Model implementation

import torch

#  Multilayer perceptron model 
# class Net(torch.jit.ScriptModule):
class Net(torch.nn.Module):
    def __init__(self, in_features):
        super(Net, self).__init__()
        self.linear_relu1 = torch.nn.Linear(in_features, 128) #  Input layer 
        self.linear_relu2 = torch.nn.Linear(128, 256) #  Hidden layer 
        self.linear_relu3 = torch.nn.Linear(256, 256) #  Hidden layer 
        self.linear_relu4 = torch.nn.Linear(256, 256) #  Hidden layer 
        self.linear5 = torch.nn.Linear(256, 1) #  Output layer 

    def forward(self, input):
        y_pred = self.linear_relu1(input)
        y_pred = torch.nn.functional.relu(y_pred)

        y_pred = self.linear_relu2(y_pred)
        y_pred = torch.nn.functional.relu(y_pred)

        y_pred = self.linear_relu3(y_pred)
        y_pred = torch.nn.functional.relu(y_pred)

        y_pred = self.linear_relu4(y_pred)
        y_pred = torch.nn.functional.relu(y_pred)

        y_pred = self.linear5(y_pred)
        return y_pred

#  Model examples . Characteristic number ：331
my_model = Net(331)

scripted_model = torch.jit.script(my_model)
print(scripted_model)
print(scripted_model.code)

#  Load the parameters of pre training 
scripted_model.load_state_dict(torch.load('model_param'))
#  Enter prediction mode 
scripted_model.eval()

#  Model serialization 
scripted_model.save("model.pt")

• C++ Call model

#include <torch/script.h>
#include <iostream>
using namespace std;

int main(int argc, const char* argv[])
{
    if (argc != 2) {
        cerr << "usage: example_app <path-to-exported-script-module>" << endl;
        return -1;
    }

    auto model = torch::jit::load(argv[1]); // torch::jit::Module

    cout << "load OK!" << endl;
    return 0;
}

• CMake

Create the following CMakeLists.txt file

cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
project(example)

find_package(Torch REQUIRED)

message(STATUS "TORCH_LIBRARIES = ${TORCH_LIBRARIES}")

set(mkl_include /opt/intel/oneapi/mkl/2022.0.0/include)
set(mkl_lib /opt/intel/oneapi/mkl/2022.0.0/lib)
include_directories(${mkl_include})
link_directories(${mkl_lib})

add_executable(example_app example_app.cpp)
target_link_libraries(example_app "${TORCH_LIBRARIES}" libmkl_intel_ilp64.dylib)
set_property(TARGET example_app PROPERTY CXX_STANDARD 14)

mkl There are still problems with the use of , To be amended ...

reference ：

• 《 Hands-on deep learning 》 — Hands-on deep learning 2.0.0-beta0 documentation
• actual combat ： Use PyTorch Building neural network to forecast house price - You know
• utilize C++ call PyTorch Model of _MLer-CSDN Blog _c++ call pytorch Model