ngboost
lleaves
ngboost | lleaves | |
---|---|---|
1 | 4 | |
1,587 | 300 | |
1.0% | - | |
6.7 | 6.7 | |
3 months ago | about 1 month ago | |
Python | Python | |
Apache License 2.0 | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
ngboost
-
Exploring Estimations and Confidence Intervals of Three Point Shooting
For DARKO, I do this by passing my time-decay/padding/kalman calculations as features (along with other pieces of biographical information) into NGboost. This gives me a point-estimate prediction, as well as confidence intervals. NGboost is somewhat limited in terms of what distributions it can handle, so you need to use care here, but it's very powerful since it can easily work with a large feature space (DARKO uses hundreds of features).
lleaves
- LLeaves: A LLVM-based compiler for LightGBM decision trees
-
Cold Showers
I built this decision tree (LightGBM) compiler last summer: https://github.com/siboehm/lleaves
It get's you ~10x speedups for batch predictions, more if your model is big. It's not complicated, it ended up being <1K lines of Python code. I heard a couple of stories like yours, where people had multi-node spark clusters running LightGBM, and it always amused me because by if you compiled the trees instead you could get rid of the whole cluster.
-
Tree compiler that speeds up LightGBM model inference by ~30x
In a near-future version I'll expose some of the compilation parameters, I was somewhat afraid of having an API that's too complicated deterring people who just want a no-fuzz drop-in replacement for LightGBM. But as long as I keep sane defaults and have the parameters optional it should be fine. Relevant parameters are definitely block size (needs to adjust to L1i size and tree size) as well as the LLVM codemodel (a smaller adress space increases single-batch prediction speeds but doesn't work for large models). The thread-size specific compilation I'm still looking into, it makes the API more complicated and so might not be worth it.
What are some alternatives?
PaddlePaddle - PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice (『飞桨』核心框架,深度学习&机器学习高性能单机、分布式训练和跨平台部署)
mljar-supervised - Python package for AutoML on Tabular Data with Feature Engineering, Hyper-Parameters Tuning, Explanations and Automatic Documentation
Keras - Deep Learning for humans
m2cgen - Transform ML models into a native code (Java, C, Python, Go, JavaScript, Visual Basic, C#, R, PowerShell, PHP, Dart, Haskell, Ruby, F#, Rust) with zero dependencies
natural-posterior-network - Official Implementation of "Natural Posterior Network: Deep Bayesian Predictive Uncertainty for Exponential Family Distributions" (ICLR, 2022)
miceforest - Multiple Imputation with LightGBM in Python
data-science-ipython-notebooks - Data science Python notebooks: Deep learning (TensorFlow, Theano, Caffe, Keras), scikit-learn, Kaggle, big data (Spark, Hadoop MapReduce, HDFS), matplotlib, pandas, NumPy, SciPy, Python essentials, AWS, and various command lines.
catboost - A fast, scalable, high performance Gradient Boosting on Decision Trees library, used for ranking, classification, regression and other machine learning tasks for Python, R, Java, C++. Supports computation on CPU and GPU.
scikit-learn - scikit-learn: machine learning in Python
spaCy - 💫 Industrial-strength Natural Language Processing (NLP) in Python