Mengapa xgboost jauh lebih cepat daripada sklearn GradientBoostingClassifier?

Saya mencoba untuk melatih model peningkatan gradien lebih dari 50k contoh dengan 100 fitur numerik. XGBClassifiermenangani 500 pohon dalam waktu 43 detik pada mesin saya, sementara GradientBoostingClassifierhanya menangani 10 pohon (!) dalam 1 menit dan 2 detik :( Saya tidak repot-repot mencoba...

scikit-learn xgboost gbm data-mining classification data-cleaning machine-learning reinforcement-learning data-mining bigdata dataset nlp language-model stanford-nlp machine-learning neural-network deep-learning randomized-algorithms machine-learning beginner career xgboost loss-function neural-network software-recommendation naive-bayes-classifier classification scikit-learn feature-selection r random-forest cross-validation data-mining python scikit-learn random-forest churn python clustering k-means machine-learning nlp sentiment-analysis machine-learning programming python scikit-learn nltk gensim visualization data csv neural-network deep-learning descriptive-statistics machine-learning supervised-learning text-mining orange data parameter-estimation python pandas scraping r clustering k-means unsupervised-learning