linux下c++集成已训练LightGBM模型进行预测

0 原理

通过下载lightGBM源码，使用cmake3.18版本对源码进行编译，生成lib_lightgbm.so文件。在c++文件中，通过链接已经编译好的lgb库，实现在c++中调用已经训练好的lgb模型。

1 相关材料

1.1 安装包

（1）lightGBM源码 提取码:qab2
（2）cmake 提取码:3sdn
（3）环境：centos7

1.2 前提准备

操作系统已经按照g++和gcc编译工具，可使用命令进行安装
yum install gcc
yum install gcc-c++

2 过程

本次操作目录均在/root/test,如下图所示:

2.1 解压安装cmake

(1) 输入命令:tar -zxvf cmake-3.18.4.tar.gz 解压cmake.tar.gz
(2) 进入cmake-3.18.4文件夹并执行./configure命令
按照过程中如果报了“Could not find OpenSSL. Install an OpenSSL development package or”，需要先执行yum instal openssl和yum install openssl-devel再执行./configure命令

(3) 执行命令gmake

(4) 执行命令确认cmake的版本，确认cmake是否成功安装

2.2 解压编译lightGBM

2.3 lightGBM模型生成

2.3.1 c++生成lgb模型

2.3.2 python生成lgb模型

2.4 c++集成lightGBM预测

#include <LightGBM/c_api.h>
#include <iostream>
#include <vector>

void predict(std::vector<float> &row)
{

    std::string pred_result = "";
    int temp;
    int p = 1;
    BoosterHandle handle;

    temp = LGBM_BoosterCreateFromModelfile("models/3_300_gbm.txt", &p, &handle);
    std::cout << "load result value is " << temp << std::endl;

    // std::vector<float> row = {0.07946399999999999, 0.9537260000000001, 0.9621209999999999, 0.976303, 7.0, 3.0};
    for (auto value : row)
        std::cout << value << ",";
    std::cout << std::endl;
    void *in_p = static_cast<void *>(row.data());
    std::vector<double> out(1, 0);
    double *out_result = static_cast<double *>(out.data());
    int64_t out_len;
    int res = LGBM_BoosterPredictForMat(handle, in_p, C_API_DTYPE_FLOAT32, 1, 6, 1, C_API_PREDICT_NORMAL, 0, -1, "None", &out_len, out_result);
    std::cout << "file predict result is:" << res << std::endl;
    std::cout << "row predict result size is " << out.size() << " value is " << out[0] << std::endl;
}

int main()
{
    std::vector<float> row = {0.07946399999999999, 0.9537260000000001, 0.9621209999999999, 0.976303, 7.0, 3.0};
    predict(row);
    std::cout << std::endl;

    std::vector<float> row1 = {0.910457, 0.692459, 0.8338110000000001, 0.78886, 14.0, 10.0};
    predict(row1);

    std::cout << "Ok complete!" << std::endl;
    return 0;
}
// g++ -g -Wall -std=c++11 test.cpp -l_lightgbm -Wl,-R /usr/local/lib -o test
// g++ -g -Wall -std=c++11 test.cpp -l_lightgbm -Wl,-R /root/moead/models -L/ydq/moead/models -I/root/moead/models/include -o test

3 参考

1 源码安装CMake3.18.4
2 如何在C++程序中调用lightgbm (How to use lightgbm in C++ program)
3 Lightgbm C++接口使用

本文地址：https://blog.csdn.net/guazhi8853/article/details/109544552