Group labels are now passed as a first column in the feature matrix to the estimator. The estimator strips this column, reorders the data and calculates the group sizes. Also this PR includes some ranking metrics that can be used to evaluate prediction quality (because SKlearns ndcg metric has been removed).
Pull request that resolves #6
Group labels are now passed as a first column in the feature matrix to the estimator. The estimator strips this column, reorders the data and calculates the group sizes. Also this PR includes some ranking metrics that can be used to evaluate prediction quality (because SKlearns ndcg metric has been removed).
Feel free to squash my commits.