The ever increasing application of machine learning models in industry and academia requires tools which are easy to use and ensure a reliable model fitting process. The R package universe covers practically all statistical models on the planet including all relevant machine learning models like neural nets, support vector machines, decision trees, and random forests. However, most of these packages do not provide a consistent interface, which makes it hard to fit and compare models from different families. Even worse, it is hard to create standardized workflows for typical machine learning projects which ensure that
- no information has been leaked from the training data, leading to higher performance numbers.
- models are compared on the same re-sampling procedures.
- performance metrics are calculated correctly.
Each lesson in the Machine Learning with Tidymodels course module covers one essential skill which together completes the entire machine learning workflow:
- The tidymodels Machine Learning Workflow: Start your machine learning journey and learn the most fundamental building blocks of the tidymodels framework.
- Data Preprocessing with recipes: Learn why data preprocessing is crucial in your machine learning workflow and create your first data transformations with the recipes package.
- Model Fitting with parsnip: Fit machine learning models using the parsnip package including linear regression, decision trees and boosting trees.
- Model Evaluation and Performance Metrics with yardstick: Estimate model quality based on different performance metrics using the yardstick package.
- Resampling techniques using rsample: Avoid overfitting by using resampling techniques including cross-validation and bootstrap using the rsample package.
- Model optimization using tune: Optimize your model parameters using the tune package to find models which predict new data well.
Get Your Personalized Cheat Sheets
With the latest update on our course platform you can create your own personalized cheat-sheets based on your progress. See also this blog post for more information.