I was doing a 3-month internship from September to November at Miltenyi Biotec. The topic of my internship was to develop a pipeline of training a machine learning model to predict different cell-types from flow cytometry data.
During the internship, I worked with a team following agile principle using Jira and Bitbucket to organize our sprints, tasks, and code. The framework which I was using called Tidymodels. Tidymodels is a collection bundle of R packages which follow the tidyverse principles that can help user to implement for machine learning tasks.
As describe above, there are multiple packages collected in Tidymodels. (1) rsample provides functions for you so sample your data. (2) parsnip helps you to set up the engine of your machine learning model. (3) recipes can prepare the steps and process you want to include in your analysis. (4) tune offers you different ways to tune hyperparameters. (5) yardstick includes function for evaluating your model performance.
workflows (or workflowsets if you have multiple models want to run) can structure your codes and perform all the analysis at once.
The GET STARTED page is a nice place to get a glance to use Tidymodels framework for machine learning task. If your are familiar with R and want to implement machine learning project with it, I will recommend you to try it out.
References: