SlickML is an open-source machine learning library written in Python aimed at accelerating the experimentation time for ML applications with tabular data while maximizing the amount of information can be inferred. Data Scientists' tasks can often be repetitive such as feature selection, model tuning, or evaluating metrics for classification and regression problems. We strongly believe that a good portion of the tasks based on tabular data can be addressed via gradient boosting and generalized linear models1. SlickML provides Data Scientists with a toolbox to quickly prototype solutions for a given problem with minimal code while maximizing the amount of information that can be inferred. Additionally, the prototype solutions can be easily promoted and served in production with our recommended recipes via various model serving frameworks including ZenML, BentoML, and Prefect. More details coming soon 🤞 ...
✨ The API documentation is available at docs.slickml.com.
To begin with, install Python version >=3.8,<3.12 and to install the library from PyPI simply run 🏃♀️ :
pip install slickml
or if you are a python poetry user, simply run 🏃♀️ :
poetry add slickml
📣 Please note that a working Fortran Compiler (gfortran) is also required to build the package. If you do not have gcc installed, the following commands depending on your operating system will take care of this requirement.
# Mac Users
brew install gcc
# Linux Users
sudo apt install build-essential gfortran
The SlickML CLI tool behaves similarly to many other CLIs for basic features. In order to find out which version of SlickML you are running, simply run 🏃♀️ :
slickml --version
In order to avoid any potential conflicts with other installed Python packages, it is
recommended to use a virtual environment, e.g. python poetry, python virtualenv, pyenv virtualenv, or conda environment. Our recommendation is to use python-poetry 🥰 for everything 😁.
✅ An example to quickly run a Feature Selection pipeline with embedded Cross-Validation and Feature-Importance visualization:
from slickml.feautre_selection import XGBoostFeatureSelector
xfs = XGBoostFeatureSelector()
xfs.fit(X, y)
xfs.plot_cv_results()
xfs.plot_frequency()
✅ An example to quickly find the tuned hyper-parameter with Bayesian Optimization:
from slickml.optimization import XGBoostBayesianOptimizer
xbo = XGBoostBayesianOptimizer()
xbo.fit(X_train, y_train)
best_params = xbo.get_best_params()
best_params
{"colsample_bytree": 0.8213916662259918,
"gamma": 1.0,
"learning_rate": 0.23148232373451072,
"max_depth": 4,
"min_child_weight": 5.632602921054691,
"reg_alpha": 1.0,
"reg_lambda": 0.39468801734425263,
"subsample": 1.0
}✅ An example to quickly train/validate a XGBoostCV Classifier with Cross-Validation, Feature-Importance, and Shap visualizations:
from slickml.classification import XGBoostCVClassifier
clf = XGBoostCVClassifier(params=best_params)
clf.fit(X_train, y_train)
y_pred_proba = clf.predict_proba(X_test)
clf.plot_cv_results()
clf.plot_feature_importance()
clf.plot_shap_summary(plot_type="violin")
clf.plot_shap_summary(plot_type="layered_violin", layered_violin_max_num_bins=5)
clf.plot_shap_waterfall()
✅ An example to train/validate a GLMNetCV Classifier with Cross-Validation and Coefficients visualizations:
from slickml.classification import GLMNetCVClassifier
clf = GLMNetCVClassifier(alpha=0.3, n_splits=4, metric="auc")
clf.fit(X_train, y_train)
y_pred_proba = clf.predict_proba(X_test)
clf.plot_cv_results()
clf.plot_coeff_path()
✅ An example to quickly visualize the binary classification metrics based on multiple thresholds:
from slickml.metrics import BinaryClassificationMetrics
clf_metrics = BinaryClassificationMetrics(y_test, y_pred_proba)
clf_metrics.plot()
✅ An example to quickly visualize some regression metrics:
from slickml.metrics import RegressionMetrics
reg_metrics = RegressionMetrics(y_test, y_pred)
reg_metrics.plot()
You can find the details of the development process in our Contributing guidelines. We strongly believe that reading and following these guidelines will help us make the contribution process easy and effective for everyone involved 🚀🌙 . Special thanks to all of our amazing contributors 👇
Please join our Slack Channel to interact directly with the core team and our small community. This is a good place to discuss your questions and ideas or in general ask for help 👨👩👧 👫 👨👩👦 .
If you use SlickML in an academic work 📃 🧪 🧬 , please consider citing it 🙏 .
@software{slickml2020,
title={SlickML: Slick Machine Learning in Python},
author={Tahmassebi, Amirhessam and Smith, Trace},
url={https://github.com/slickml/slick-ml},
version={0.2.0},
year={2021},
}
@article{tahmassebi2021slickml,
title={Slickml: Slick machine learning in python},
author={Tahmassebi, Amirhessam and Smith, Trace},
journal={URL available at: https://github. com/slickml/slick-ml},
year={2021}
}