Wines of the PNW

DATA 505 HW #2 Kayle Megginson February 2nd 2025

Abstract:

This is a technical blog post of both an HTML file and .qmd file hosted on GitHub pages.

Setup

Step Up Code:

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(caret)

Loading required package: lattice

Attaching package: 'caret'

The following object is masked from 'package:purrr':

    lift

library(fastDummies)
wine <- readRDS(gzcon(url("https://github.com/cd-public/D505/raw/master/dat/wine.rds")))

Explanataion:
Line 1: library(tidyverse): Loads the Tidyverse, a collection of R packages for data manipulation and visualization.
Line 2: library(caret): Loads the Caret package for machine learning tasks.
Line 3: library(fastDummies): Loads the fastDummies package to create dummy variables efficiently.
Line 4: wine <- readRDS(...): Reads in the wine.rds dataset from a remote GitHub repository.

Feature Engineering

We begin by engineering an number of features.

1. Create a total of 10 features (including points).
2. Remove all rows with a missing value.
3. Ensure only log(price) and engineering features are the only columns that remain in the wino dataframe.

wino <- wine %>% 
  mutate(
    lprice = log(price),
    vintage = year,
    is_us_wine = ifelse(country == "US", 1, 0),
    name_length = nchar(title),
    description_length = nchar(description),
    region_count = (!is.na(region_1)) + (!is.na(region_2)),
    has_designation = ifelse(!is.na(designation), 1, 0),
    has_taster = ifelse(!is.na(taster_name), 1, 0),
    price_per_point = price / (points + 1),
    province_count = as.integer(as.factor(province))
  ) %>% 
  select(lprice, points, vintage, is_us_wine, name_length, description_length, 
         region_count, has_designation, has_taster, price_per_point, province_count) %>% 
  drop_na()

Caret

We now use a train/test split to evaluate the features.

1. Use the Caret library to partition the wino dataframe into an 80/20 split.
2. Run a linear regression with bootstrap resampling.
3. Report RMSE on the test partition of the data.

# Partition data into 80% training and 20% testing
set.seed(42)
train_index <- createDataPartition(wino$lprice, p = 0.8, list = FALSE)
train_data <- wino[train_index, ]
test_data <- wino[-train_index, ]

# Train a linear regression model
fit_control <- trainControl(method = "boot", number = 50)
model <- train(lprice ~ ., data = train_data, method = "lm", trControl = fit_control)

# Predict on test data and calculate RMSE
predictions <- predict(model, test_data)
rmse_value <- RMSE(predictions, test_data$lprice)

rmse_value

[1] 0.3892879

We recieve an output of 0.3892879 as our RMSE value

Variable selection

We now graph the importance of your 10 features.

# Extract feature importance
importance <- varImp(model)

# Plot feature importance
ggplot(importance) +
  ggtitle("Feature Importance in Predicting Log(Price)")