In order to use protein sequences as inputs for deep learning models they need to be transformed into numerical descriptors. In this notebook I highlight four different ways to generate such numerical features from protein sequences.
In this post I explain how I build and tuned a Machine Learning model in R using the tidymodels package to predict passenger survival on the Titanic. I define some new features from the data to enhance the accuracy of the model and perform tuning of the model in Python using optuna. The final model reaches a top 2% score on Kaggle.
In this post I explain how I build and tuned a Machine Learning model in R using the tidymodels package. I then show how to tune the hyperparameters directly in R using a grid search and compare the results with performing the tuning in optuna using the reticulate package.
An important step in building ML models is choosing a set of optimal hyperparameters. One of the best python packages to tune hyperparameters is Optuna. In this article I use Optuna to estimate the minimum and maximum of a simple function and try to illustrate, with a set of interactive graphs, what makes Optuna so efficient.