the meaning and cause of "[LightGBM] [Fatal] Check failed: (train_data->num_features()) > (0)"

[royshan]

Hi,

When working with a high-dimension dataset which has many features but small samples, I often run into the following error: [LightGBM] [Fatal] Check failed: (train_data->num_features()) > (0) at c:\users\builder\appdata\local\temp\pip-req-build-zpa7v636\compile\src\treelearner\feature_histogram.hpp, line 1218 .

That line is in the function of void ResetConfig(const Dataset* train_data, const Config* config).

I don't understand why ResetConfig was called. What's the meaning and possible causes? Does that suggest I should abandon the model at all and adjust hyper-parameters?

Thank you for your help!

[royshan]

Any comment or suggestion?