Revision history [back]

Certainly! Here is an example of how to use LightGBM in C++ with Microsoft Visual Studio:

1. First, download and install LightGBM from the official website.
2. Create a new Visual Studio project.
3. In the project properties, add the path to the LightGBM include directory to the "Include Directories" field.
4. In the project properties, add the path to the LightGBM library directory to the "Library Directories" field.
5. In the project properties, add "lightgbm.lib" to the "Additional Dependencies" field.

6. Include the LightGBM header file in your main .cpp file:

   #include <LightGBM/lightgbm.h>

7. Initialize a Dataset object with your training data:

   LightGBM::Dataset train_data = LightGBM::Dataset::CreateFromFile(
       "train.txt", 
       "", 
       0, 
       0, 
       true, 
       "", 
       "1", 
       1, 
       nullptr
   );
   

8. Define a parameter object and set the necessary parameters:

   LightGBM::Config config;
   config.num_iterations = 100;
   config.boosting_type = LightGBM::BoostingType::kGBDT;
   config.max_depth = 6;
   config.learning_rate = 0.1f;
   

9. Train the model using the LightGBM object:

   LightGBM::LightGBM model(config);
   model.Train(&train_data);    
   

10. Make predictions using the trained model:

    std::vector<float> predict_data = {1.0f, 2.0f, 3.0f};
    std::vector<float> predictions;
    model.Predict(predict_data.data(), &predictions);
    

And that's it! You now have a working example of how to use LightGBM in C++ with Microsoft Visual Studio.