Use tidymodels instead of base libraries

Certainly! The **tidymodels** framework in R provides a cohesive set of packages designed for modeling and machine learning. It wraps many base R functions into a user-friendly syntax and structure for building and evaluating models. Here’s a simple example of how you can use **tidymodels** to create a machine learning pipeline. In this example, we will use the **iris** dataset to build a model predicting species based on features. We'll follow these steps: 1. Load necessary libraries. 2. Split the data into training and testing sets. 3. Preprocess the data. 4. Train a model (using logistic regression as an example). 5. Evaluate the model. Here’s a complete example: ```R # Load necessary libraries library(tidymodels) # Set the seed for reproducibility set.seed(123) # Load the iris dataset data("iris") # Split the data into training (80%) and testing (20%) sets iris_split <- initial_split(iris, prop = 0.8, strata = Species) iris_train <- training(iris_split) iris_test <- testing(iris_split) # Define a recipe for preprocessing # We'll center and scale the numeric variables iris_recipe <- recipe(Species ~ ., data = iris_train) %>% step_normalize(all_numeric_predictors()) %>% step_dummy(all_nominal_predictors(), one_hot = TRUE) # Specify the model (using logistic regression here) logistic_model <- logistic_reg() %>% set_engine("glm") # Create a workflow to tie the recipe and model together iris_workflow <- workflow() %>% add_recipe(iris_recipe) %>% add_model(logistic_model) # Fit the model on the training set iris_fit <- fit(iris_workflow, data = iris_train) # Make predictions on the test set iris_predictions <- predict(iris_fit, iris_test, type = "class") %>% bind_cols(iris_test) # Evaluate the model's performance iris_metrics <- iris_predictions %>% metrics(truth = Species, estimate = .pred_class) # Show the metrics print(iris_metrics) # Confusion matrix for further evaluation confusion_matrix <- iris_predictions %>% conf_mat(truth = Species, estimate = .pred_class) # Print the confusion matrix print(confusion_matrix) ``` ### Explanation of the Code: 1. **Loading Libraries**: We first load the `tidymodels` library, which includes several packages for data wrangling, modeling, and evaluation. 2. **Data Splitting**: We split the iris dataset into training and testing datasets while stratifying by species to ensure balanced splits. 3. **Recipe**: We create a preprocessing recipe that normalizes numeric predictors and converts categorical predictors into dummy variables. 4. **Model Specification**: We specify a logistic regression model using the `logistic_reg()` function and set the engine to "glm". 5. **Workflow**: We create a workflow by combining the recipe and model. 6. **Model Training**: We fit the model to the training dataset using the `fit()` function. 7. **Making Predictions**: We predict the species for the test dataset. 8. **Evaluation**: We compute various metrics like accuracy, and also produce a confusion matrix for detailed evaluation of model predictions. ### Notes: - This is just a simple example, and there are various hyperparameter options, different models, and preprocessing techniques available in **tidymodels**. Make sure that you have `tidymodels` installed in your R environment. You can install it by running: ```R install.packages("tidymodels") ``` You may need to install additional packages used in the recipe or modeling steps, which can be handled automatically by **tidymodels**.