Summary  
This chapter demonstrates how to specify, fit, and interpret a linear regression model using the tidymodels parsnip interface in R, including extracting model coefficients and performance metrics.  

General domain of usage  
Real estate price prediction

Regression modeling allows you to predict a continuous outcome based on one or more input features, making it an essential tool for exploring relationships between variables and making informed predictions. In business, economics, health sciences, and many other fields, regression models help you estimate values such as house prices, sales, or patient outcomes based on relevant predictors. **Tidymodels** provides a unified and tidy approach for building, fitting, and interpreting regression models in R, streamlining the process from model specification to extracting actionable results.

options(crayon.enabled = FALSE)
library(tidymodels)

# Load example housing dataset
data("ames", package = "modeldata")

# Split data into training and testing sets
set.seed(123)
housing_split <- initial_split(ames, prop = 0.8)
housing_train <- training(housing_split)
housing_test <- testing(housing_split)

# Specify a linear regression model using parsnip
lm_spec <- linear_reg() %>%
  set_engine("lm")

# Fit the model to the training data
lm_fit <- lm_spec %>%
  fit(Sale_Price ~ Gr_Liv_Area + Year_Built, data = housing_train)

# Extract model summary and coefficients
summary(lm_fit$fit)
coef(lm_fit$fit)

When building a regression model with Tidymodels, you begin by specifying the model type and engine using the `parsnip` package. Here, `linear_reg()` defines a linear regression model, and `set_engine("lm")` selects R's base linear modeling engine. You then fit the model to your data using the `fit()` function, providing a formula that describes the relationship between the outcome and predictors, along with the training dataset. Once fitted, you can extract results such as model summaries and coefficients directly from the fitted model object. These outputs allow you to interpret the influence of each predictor on the outcome and assess model performance.

library(testthat)

source("user_code.R")
test_that("fit_linear_model returns a fitted parsnip model object", {
    expect_true(exists("fit_linear_model"),
        info = "fit_linear_model function should be defined."
    )
    res <- fit_linear_model(df)
    expect_true(inherits(res, "model_fit"),
        info = "Returned object should be a fitted linear regression model from parsnip."
    )
    expect_true(!is.null(res$fit), info = "Fitted model should contain a non-null fit component.")
})

test_that("fit_linear_model uses y as outcome and x1, x2 as predictors", {
    res <- fit_linear_model(df)
    formula_used <- formula(res$fit)
    expect_equal(formula_used, y ~ x1 + x2,
        info = "Model formula should use y as outcome and x1, x2 as predictors."
    )
})

test_that("fit_linear_model works for provided sample dataset", {
    expect_error(fit_linear_model(df), NA,
        info = "Function should work without error for the sample dataset."
    )
})


test_main.R

Master predictive modeling in R using the Tidymodels framework. This course guides you through the entire modeling workflow, from data preprocessing to model evaluation and interpretation, leveraging the power of the tidyverse and modern machine learning techniques.

Comprehensive journey through predictive modeling using Tidymodels in R, covering the entire workflow from data preparation to model deployment.

Building Regression Models

Oplossing