[R-package] exclude training data from being checked for early stopping

[StrikerRUS]

Check all datasets in validation list for early stopping excluding the training data to match the Python-package behavior:

Early stopping requires at least one set in valid_sets. If there is more than one, it will use all of them except the training data:
https://github.com/microsoft/LightGBM/blob/master/docs/Python-Intro.rst#early-stopping

Refer to #2371.
Python code:

LightGBM/python-package/lightgbm/callback.py

Lines 238 to 241 in 5c0daca

	if ((env.evaluation_result_list[i][0] == "cv_agg" and eval_name_splitted[0] == "train"
	or env.evaluation_result_list[i][0] == env.model._train_data_name)):
	_final_iteration_check(env, eval_name_splitted, i)
	continue # train data for lgb.cv or sklearn wrapper (underlying lgb.train)

C++ code:

LightGBM/src/io/config.cpp

Lines 237 to 248 in 5c0daca

	if (valid_data_initscores.empty()) {
	std::vector<std::string> new_valid;
	for (size_t i = 0; i < valid.size(); ++i) {
	if (valid[i] != data) {
	// Only push the non-training data
	new_valid.push_back(valid[i]);
	} else {
	is_provide_training_metric = true;
	}
	}
	valid = new_valid;
	}

[StrikerRUS]

Closed in favor of being in #2302. We decided to keep all feature requests in one place.

Welcome to contribute this feature! Please re-open this issue (or post a comment if you are not a topic starter) if you are actively working on implementing this feature.

[jameslamb]

this was fixed in #2961