29 Machine Learning Models

• Both R and Python include several options for implementing machine learning models

• Python ecosystem of packages to fit machine learning models are more consistent

• Several packages in R are available to call Python libraries from R environment

---

29.1 R: Machine Learning Models

• Statisticians and modellers developed several useful packages to fit machine learning models in both supervised and unsupervised framework

• Some comprehensive R packages for machine learning models are: nnet, RWeka, randomForest, ranger, e1071, kernlab, gbm, xgboost, glmnet

• Packages tensorflow provides an interface to tensorflow and torch provides an interface to PyTorch

• Packages that provide miscellaneous functions to build predictive models: caret, mlr3, tidymodels, h2o

• In this workshop, we will not provide further implementation of machine learning models in R

• Please check the corresponding package websites for further information

---

29.2 Python: Machine Learning Models

• Python package scikit-learn fits a comprehensive range of machine learning models

• scikit-learn: Machine Learning in Python

• The package is built on other packages like NumPy, SciPy and Matplotlib

• It includes modelling option for both supervised and unsupervised problems

• Methodologies include classification, regression, clustering, dimension reduction

• It also includes several supportive tools likes model selection, preprocessing, feature extraction etc.

• Install scikit-learn and its dependencies from PyPI repository: pip install scikit-learn

---