# Assuming 'fastfood' is your dataset # Step 1: Filter the data to include only restaurants with between 50 and 60 items filtered_data % group_by(restaurant)

# Assuming 'fastfood' is your dataset # Step 1: Filter the data to include only restaurants with between 50 and 60 items filtered_data <- fastfood %>% group_by(restaurant) %>% filter(n() >= 50 & n() <= 60) %>% ungroup() # Step 2: Ensure 'restaurant' has more than one level and convert to factor if necessary filtered_data <- filtered_data %>% mutate(restaurant = as.factor(restaurant)) # If the 'restaurant' variable has fewer than 2 levels, exclude it from the model if (nlevels(filtered_data$restaurant) <= 1) { predictors <- c("cholesterol", "total_carb", "vit_a") } else { predictors <- c("cholesterol", "total_carb", "vit_a", "restaurant") } # Step 3: Run the initial regression model with the appropriate predictors formula <- as.formula(paste("total_fat ~", paste(predictors, collapse = " + "))) initial_model <- lm(formula, data = filtered_data) # Step 4: Identify significant predictors significant_predictors <- tidy(initial_model) %>% filter(p.value <= 0.05) %>% pull(term) %>% setdiff("(Intercept)") # If no predictors are significant, raise an error if (length(significant_predictors) == 0) { stop("No significant predictors found in the initial model.") } # Step 5: Create formula with only significant predictors for the final model final_formula <- as.formula(paste("total_fat ~", paste(significant_predictors, collapse = " + "))) # Re-run the model with significant predictors only final_model <- lm(final_formula, data = filtered_data