Algorithms

Right now there are available algorithms for following machine learning tasks:

  • binary classification
  • regression

More is coming soon!

Binary Classification

For binary classification there are following algorithms available:

  • Xgboost
  • LightGBM
  • Random Forest
  • Extra Trees
  • Regularized Greedy Forest
  • k-Nearest Neighbors
  • Logistic Regression
  • Neural Networks

Regression

For regression task there are following algorithms available:

  • Xgboost
  • LightGBM
  • Random Forest
  • Extra Trees
  • Regularized Greedy Forest

Libraries

The MLJAR is standing on the shoulders of giants:

Ensemble

There is an option to build ensemble of models based on trained algorithms. Currently, there is available an ensemble average method, which does a greedy search over all results and try to add (with repetition) a model to the ensemble to improve ensemble performance. The ensemble performance is computed based on out-of-folds predictions of used models. For more details please take a look on Caruana paper.