MLSolver

MLSolver is an open source AutoML framework. In the future, the idea is to be able to handle tabular data problems (normal, time series), computer vision problems and NLP problems.

TODO

Normal Tabular Data Regression and Classification

* Hyperparameter tuning with Optuna

* Add OOP patters where it is needed

* Improve pipeline parameters

* Fill nan strategies
* Feature selection strategies
* ...

* Feature engineering

* Create feature combinations with operators +-*/
* Create bins using quantiles
* Create cluster features

* Add more metrics determining if needs to be maximized and minimized

* Improve folds package

* Handle different regression and classification types of target

* Single column regression
* Multiple column regression
* Binary classification
* Multiclass classification
* Multilabel classification

* Add more models

* Neural Networks
* Catboost
* Sklearn Gradient Boosting algorithms

* Create ensemble models

* Bagging
* Blending
* Stacking

* Testing

* Create demos for end-to-end testing
* Test functions

Time Series Tabular Data Regression and Classification

Computer Vision

Natural Language Processing