"one-hot encoding can seriously degrade tree-model performance" Why it's important to use an RF/GBM that honors categorical vars (e.g. H2O) x.com/h2oai/status/8…

“Exact xgboost and Fast Histogram xgboost training speed comparison” by Laurae medium.com/data-design/ex…

Does anyone have experience with gradient boosting from SMILE github.com/haifengl/smile ?

Missed our meetup last night w/ Arno Candel on #gpu accelerated #xgboost? Here's the recording: youtu.be/_kwdeYGHO8g Happy viewing!

GBM math and scikit learn code walkthrough medium.com/towards-data-s… x.com/Horatio_JSY/st…

#MachineLearning techniques, (#deeplearning, #randomforest, and #XGBoost) are common at top #riskanalytics units. bit.ly/2s4AsbG

Slides+video of my #useR2017 talk on #machinelearning tools with H2O.ai #lightgbm #xgboost #GPU #deeplearning channel9.msdn.com/Events/useR-in…

I changed my tw handle to @SzilardPafka Despite #deeplearning hype #GBM delivers best in class accuracy #MachineLearning #algorithms H2O.ai

Boosted decision tree benchmarks: XGBoost vs LightGBM with GPUs blogs.technet.microsoft.com/machinelearnin…

Gradient Boosting explained by Anton Rogozhnikov buff.ly/2uL5MMK #MachineLearning

#LightGBM vs #XGboost benchmarks Microsoft blogs.technet.microsoft.com/machinelearnin… and I changed my tw handle to @SzilardPafka github.com/szilard/GBM-pe… find LightGBM at least faster

#LightGBM will be faster than #XGBoost but AUC result will depend on your data. Give it a try. All benchmarks are wrong, but some are useful

Use quantile-based binning of numeric variables when dealing with data that have a lot of outliers! (2/2)

Well done! 4th and 6th places in Instacart Market Basket Analysis Kaggle competition achieved with CatBoost, kaggle.com/c/instacart-ma…

Great overview of ensemble methods blog.statsbot.co/ensemble-learn… from Statsbot. See how #stacking relates to #randomforest and #boosting.

Does feature engineering matter when doing Random Forest or Gradient Boosting? stats.stackexchange.com/questions/3002…

New #GradientBoosting implementation in TensorFlow is designed to work on data that do not fit into memory arxiv.org/abs/1710.11555…

Classification tree (#xgboost) does not identify perfectly seperable data stats.stackexchange.com/q/307534?stw=2

How does xgboost select which feature to split on? stats.stackexchange.com/q/299859?stw=2