众所周知 Python 做机器学习/深度学习非常牛,一些为人熟知的框架:scikit-learn、TensorFlow、keras、PyTorch 。那 R 能不能做?强不强?
我最初是从张敬信老师那了解到tidyverse
,知道了 R 数据处理的强大(R | R入门の想法 + tidyverse)。现在又从他那知道了 R 做机器学习的强大(mlr3verse、tidymodels)。
具体请看:
知乎 | 张敬信 | 【R语言】最新机器学习框架与学习资料推荐
知乎 | 张敬信 | 【R机器学习】mlr3verse vs tidymodels
# 以上收录于张敬信老师的知乎专栏 R机器学习:基于mlr3verse
知乎 | 张敬信 | 【R语言】新一代机器学习包mlr3快速入门
此外:
- tidymodels
https://www.tmwr.org/
https://github.com/tidymodels/tidymodels
https://www.tidymodels.org/(有知友 capas 翻译)
https://tidymodels.tidymodels.org/(须翻墙)
知乎 | 阿道克 | tidymodels初探(上)
知乎 | 阿道克 | tidymodels初探(下)
- mlr3
https://mlr3book.mlr-org.com/ (须翻墙)
https://cheatsheets.mlr-org.com/mlr3.pdf
公众号 | 医学和生信笔记 | mlr3:基础使用(注意有一个合集)
王诗翔 | mlr3(一)快速入门
王诗翔 | mlr3(二)基础
王诗翔 | mlr3(三)模型优化
......
1. tidymodels
b 站 PartTimeAlterEgo 这位老哥正在更新 【简单的tidymodels】系列,配合 https://rpubs.com/liam/tutorialOfTidymodels 网址。
2. mlr3verse
官网提到了“For beginners, we strongly recommend to install and load the mlr3verse package for a better user experience”。
- mlr3: Machine Learning in R - Next Generation (github)
- mlr3cluster: Unsupervised Clustering (github)
- mlr3data: Additional data sets and tasks (github)
- mlr3filters: Filter Based Feature Selection (github)
- mlr3fselect: Wrapper Based Feature Selection (github)
- mlr3learners: Recommended Learners (github)
- mlr3pipelines: Preprocessing Operators and Pipelines (github)
- mlr3proba: Probalistic Regression and Survival Analysis (github)
- mlr3tuning: Hyperparameter Tuning (github)
- mlr3viz: Visualizations: https://github.com/mlr-org/mlr3viz
- paradox: Parameter Spaces (github)
> install.packages(mlr3verse)
> library(mlr3verse)
> mlr3verse_info()
package version
1: bbotk 0.5.3
2: mlr3cluster 0.1.3
3: mlr3data 0.6.0
4: mlr3filters 0.5.0
5: mlr3fselect 0.7.1
6: mlr3hyperband 0.4.1
7: mlr3learners 0.5.3
8: mlr3misc 0.10.0
9: mlr3pipelines 0.4.1
10: mlr3tuning 0.13.1
11: mlr3tuningspaces 0.3.0
12: mlr3viz 0.5.9
13: paradox 0.9.0