Wine Features
Hannah S.K. Pahama
February 2025
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 1.9.4 ✔ 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 errorslibrary(caret)## Loading required package: lattice
##
## Attaching package: 'caret'
##
## The following object is masked from 'package:purrr':
##
## liftlibrary(fastDummies)
wine <- readRDS(gzcon(url("https://github.com/cd-public/D505/raw/master/dat/wine.rds")))Abstract
This contains my assignment for features mainly about carret.
Step Up Code
Explanation: - tidyverse - load tidyverse for data
manipulation and visualization - caret - for machine
learning functionalities - fastDummies - for creating dummy
variables if needed - wine - loads the wine dataset from
GitHub
Feature Engineering
We begin by engineering a number of features.
- Create a total of 10 features (including points).
- Remove all rows with a missing value.
- Ensure only
log(price)and engineering features are the only columns that remain in thewine_cleandataframe.
Data Cleaning
print(colnames(wine))## [1] "id" "country" "description"
## [4] "designation" "points" "price"
## [7] "province" "region_1" "region_2"
## [10] "taster_name" "taster_twitter_handle" "title"
## [13] "variety" "winery" "year"sum(is.na(wine))## [1] 104228colSums(is.na(wine))## id country description
## 0 53 0
## designation points price
## 25200 0 0
## province region_1 region_2
## 53 17844 56475
## taster_name taster_twitter_handle title
## 0 4603 0
## variety winery year
## 0 0 0wine_clean <- wine %>% select(-designation, -region_2)
wine_clean <- wine_clean %>% drop_na(country, province, region_1)
colSums(is.na(wine_clean))## id country description
## 0 0 0
## points price province
## 0 0 0
## region_1 taster_name taster_twitter_handle
## 0 0 2726
## title variety winery
## 0 0 0
## year
## 0wine_clean <- wine_clean %>% select(-taster_twitter_handle)
colSums(is.na(wine_clean))## id country description points price province
## 0 0 0 0 0 0
## region_1 taster_name title variety winery year
## 0 0 0 0 0 0wine_clean <- wine_clean %>%
mutate(
log_price = log(price),
points_squared = points^2,
price_per_point = price / points,
price_per_year = ifelse(!is.na(year), price / (2025 - year), NA),
points_price_ratio = points / price,
is_expensive = ifelse(price > 50, 1, 0),
wine_age = ifelse(!is.na(year), 2025 - year, NA),
description_length = nchar(description),
variety_type = ifelse(variety %in% c("Cabernet Sauvignon", "Merlot", "Pinot Noir"), "Red", "Other"),
country_region = paste(country, province, sep = "_")
) %>%
drop_na()
wine_clean <- wine_clean %>%
select(log_price, points_squared, price_per_point, price_per_year,
points_price_ratio, is_expensive, wine_age, description_length,
variety_type, country_region)Caret
We now use a train/test split to evaluate the features.
set.seed(123)
train_index <- createDataPartition(wine_clean$log_price, p = 0.8, list = FALSE)
train_data <- wine_clean[train_index, ]
test_data <- wine_clean[-train_index, ]
dim(train_data)## [1] 57372 10dim(test_data)## [1] 14340 10nrow(train_data) / nrow(wine_clean)## [1] 0.8000335nrow(test_data) / nrow(wine_clean)## [1] 0.1999665train_control <- trainControl(method = "boot", number = 100)
model <- train(log_price ~ ., data = wine_clean, method = "lm", trControl = train_control)## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful cases
## Warning in predict.lm(modelFit, newdata): prediction from rank-deficient fit;
## attr(*, "non-estim") has doubtful casessummary(model)##
## Call:
## lm(formula = .outcome ~ ., data = dat)
##
## Residuals:
## Min 1Q Median 3Q Max
## -8.0684 -0.0674 -0.0254 0.0518 1.8355
##
## Coefficients:
## Estimate Std. Error t value
## (Intercept) 3.141e+00 9.540e-03 329.218
## points_squared 9.679e-05 1.192e-06 81.198
## price_per_point 4.053e-01 4.824e-03 84.025
## price_per_year -1.178e-02 6.879e-04 -17.120
## points_price_ratio -2.026e-01 2.985e-04 -678.842
## is_expensive 3.308e-01 1.474e-03 224.443
## wine_age 9.612e-04 1.886e-04 5.097
## description_length 1.075e-05 8.299e-06 1.295
## variety_typeRed 2.439e-02 1.182e-03 20.643
## country_regionArgentina_Other -7.589e-03 5.518e-03 -1.375
## `country_regionAustralia_Australia Other` 1.169e-01 9.237e-03 12.654
## `country_regionAustralia_New South Wales` -2.273e-02 1.351e-02 -1.682
## `country_regionAustralia_South Australia` -2.962e-02 3.999e-03 -7.406
## country_regionAustralia_Tasmania -1.171e-02 2.165e-02 -0.541
## country_regionAustralia_Victoria -2.486e-02 7.304e-03 -3.403
## `country_regionAustralia_Western Australia` -3.823e-02 7.643e-03 -5.002
## `country_regionCanada_British Columbia` -3.718e-02 9.323e-03 -3.988
## country_regionCanada_Ontario -1.166e-02 1.255e-02 -0.929
## country_regionFrance_Alsace -4.282e-02 3.487e-03 -12.279
## country_regionFrance_Beaujolais -5.384e-02 4.517e-03 -11.919
## country_regionFrance_Bordeaux -2.010e-02 2.831e-03 -7.100
## country_regionFrance_Burgundy 2.515e-02 3.081e-03 8.163
## country_regionFrance_Champagne 5.709e-02 6.200e-03 9.208
## `country_regionFrance_France Other` 1.844e-02 6.368e-03 2.896
## `country_regionFrance_Languedoc-Roussillon` -6.374e-03 5.590e-03 -1.140
## `country_regionFrance_Loire Valley` -4.015e-02 3.765e-03 -10.664
## country_regionFrance_Provence -4.914e-02 4.408e-03 -11.148
## `country_regionFrance_Rhône Valley` -1.484e-03 4.224e-03 -0.351
## `country_regionFrance_Southwest France` -2.624e-02 3.869e-03 -6.782
## `country_regionItaly_Central Italy` -2.526e-02 5.202e-03 -4.855
## `country_regionItaly_Italy Other` 7.740e-03 1.525e-02 0.507
## country_regionItaly_Lombardy -1.090e-02 6.905e-03 -1.579
## `country_regionItaly_Northeastern Italy` -3.090e-02 3.981e-03 -7.763
## `country_regionItaly_Northwestern Italy` -7.852e-02 5.101e-02 -1.539
## country_regionItaly_Piedmont 2.179e-02 3.356e-03 6.494
## `country_regionItaly_Sicily & Sardinia` -2.106e-02 4.631e-03 -4.548
## `country_regionItaly_Southern Italy` -2.916e-02 5.336e-03 -5.465
## country_regionItaly_Tuscany -2.703e-03 2.934e-03 -0.921
## country_regionItaly_Veneto 1.667e-03 4.713e-03 0.354
## country_regionSpain_Andalucia -5.049e-02 1.694e-02 -2.981
## country_regionSpain_Catalonia -1.421e-02 4.557e-03 -3.118
## `country_regionSpain_Central Spain` 9.075e-02 5.733e-03 15.829
## country_regionSpain_Galicia -6.404e-02 6.221e-03 -10.293
## country_regionSpain_Levante 5.717e-02 5.943e-03 9.619
## `country_regionSpain_Northern Spain` 6.269e-03 2.808e-03 2.233
## `country_regionSpain_Spain Other` 8.119e-02 1.675e-02 4.846
## `country_regionSpain_Spanish Islands` -3.227e-02 2.386e-02 -1.352
## country_regionUS_Arizona -1.239e-02 3.611e-02 -0.343
## country_regionUS_California -8.751e-03 2.324e-03 -3.766
## country_regionUS_Colorado 7.316e-03 1.694e-02 0.432
## country_regionUS_Connecticut -1.495e-02 8.062e-02 -0.185
## country_regionUS_Idaho 1.551e-01 2.247e-02 6.901
## country_regionUS_Iowa -7.319e-02 1.140e-01 -0.642
## country_regionUS_Kentucky 6.041e-02 1.140e-01 0.530
## country_regionUS_Massachusetts -5.961e-02 8.062e-02 -0.739
## country_regionUS_Michigan -4.291e-03 1.155e-02 -0.372
## country_regionUS_Missouri -2.642e-02 6.584e-02 -0.401
## country_regionUS_Nevada -3.246e-02 6.584e-02 -0.493
## `country_regionUS_New Jersey` -2.395e-02 5.102e-02 -0.469
## `country_regionUS_New Mexico` -1.721e-02 2.497e-02 -0.689
## `country_regionUS_New York` -4.634e-02 3.141e-03 -14.756
## `country_regionUS_North Carolina` -4.951e-02 2.951e-02 -1.678
## country_regionUS_Ohio -1.498e-02 4.035e-02 -0.371
## country_regionUS_Oregon -2.781e-02 2.697e-03 -10.310
## country_regionUS_Pennsylvania -3.886e-02 3.169e-02 -1.226
## country_regionUS_Texas -1.608e-02 1.361e-02 -1.181
## country_regionUS_Vermont 1.457e-02 6.584e-02 0.221
## country_regionUS_Virginia -1.771e-02 6.515e-03 -2.718
## country_regionUS_Washington -1.504e-02 2.472e-03 -6.083
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## points_squared < 2e-16 ***
## price_per_point < 2e-16 ***
## price_per_year < 2e-16 ***
## points_price_ratio < 2e-16 ***
## is_expensive < 2e-16 ***
## wine_age 3.46e-07 ***
## description_length 0.195301
## variety_typeRed < 2e-16 ***
## country_regionArgentina_Other 0.169017
## `country_regionAustralia_Australia Other` < 2e-16 ***
## `country_regionAustralia_New South Wales` 0.092646 .
## `country_regionAustralia_South Australia` 1.32e-13 ***
## country_regionAustralia_Tasmania 0.588663
## country_regionAustralia_Victoria 0.000666 ***
## `country_regionAustralia_Western Australia` 5.69e-07 ***
## `country_regionCanada_British Columbia` 6.67e-05 ***
## country_regionCanada_Ontario 0.352860
## country_regionFrance_Alsace < 2e-16 ***
## country_regionFrance_Beaujolais < 2e-16 ***
## country_regionFrance_Bordeaux 1.26e-12 ***
## country_regionFrance_Burgundy 3.33e-16 ***
## country_regionFrance_Champagne < 2e-16 ***
## `country_regionFrance_France Other` 0.003781 **
## `country_regionFrance_Languedoc-Roussillon` 0.254188
## `country_regionFrance_Loire Valley` < 2e-16 ***
## country_regionFrance_Provence < 2e-16 ***
## `country_regionFrance_Rhône Valley` 0.725396
## `country_regionFrance_Southwest France` 1.20e-11 ***
## `country_regionItaly_Central Italy` 1.20e-06 ***
## `country_regionItaly_Italy Other` 0.611866
## country_regionItaly_Lombardy 0.114291
## `country_regionItaly_Northeastern Italy` 8.41e-15 ***
## `country_regionItaly_Northwestern Italy` 0.123751
## country_regionItaly_Piedmont 8.42e-11 ***
## `country_regionItaly_Sicily & Sardinia` 5.43e-06 ***
## `country_regionItaly_Southern Italy` 4.65e-08 ***
## country_regionItaly_Tuscany 0.356986
## country_regionItaly_Veneto 0.723593
## country_regionSpain_Andalucia 0.002874 **
## country_regionSpain_Catalonia 0.001820 **
## `country_regionSpain_Central Spain` < 2e-16 ***
## country_regionSpain_Galicia < 2e-16 ***
## country_regionSpain_Levante < 2e-16 ***
## `country_regionSpain_Northern Spain` 0.025557 *
## `country_regionSpain_Spain Other` 1.27e-06 ***
## `country_regionSpain_Spanish Islands` 0.176293
## country_regionUS_Arizona 0.731536
## country_regionUS_California 0.000166 ***
## country_regionUS_Colorado 0.665889
## country_regionUS_Connecticut 0.852903
## country_regionUS_Idaho 5.20e-12 ***
## country_regionUS_Iowa 0.520812
## country_regionUS_Kentucky 0.596187
## country_regionUS_Massachusetts 0.459671
## country_regionUS_Michigan 0.710177
## country_regionUS_Missouri 0.688203
## country_regionUS_Nevada 0.621954
## `country_regionUS_New Jersey` 0.638720
## `country_regionUS_New Mexico` 0.490753
## `country_regionUS_New York` < 2e-16 ***
## `country_regionUS_North Carolina` 0.093359 .
## country_regionUS_Ohio 0.710462
## country_regionUS_Oregon < 2e-16 ***
## country_regionUS_Pennsylvania 0.220110
## country_regionUS_Texas 0.237501
## country_regionUS_Vermont 0.824830
## country_regionUS_Virginia 0.006570 **
## country_regionUS_Washington 1.18e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.114 on 71643 degrees of freedom
## Multiple R-squared: 0.9691, Adjusted R-squared: 0.9691
## F-statistic: 3.305e+04 on 68 and 71643 DF, p-value: < 2.2e-16model$results## intercept RMSE Rsquared MAE RMSESD RsquaredSD MAESD
## 1 TRUE 0.1168612 0.9674893 0.07654025 0.009384021 0.005157647 0.002511029predictions <- predict(model, newdata = wine_clean)
rmse <- sqrt(mean((predictions - wine_clean$log_price)^2))
print(paste("RMSE on the full dataset:", round(rmse, 2)))## [1] "RMSE on the full dataset: 0.11"Model Performance Summary
- RMSE (Resampling): 0.11897 — Low, indicating good fit.
- R-squared (Resampling): 0.9664 — Explains 96.64% of the variance, excellent!
- MAE (Resampling): 0.0771 — Low error on average.
- RMSE on Test Data: 0.1123 — Slightly lower than resampling, showing strong generalization.
Conclusion: The model performs well with minimal error, high R-squared, and good consistency across resampling iterations.
Variable Selection
importance <- varImp(model, scale = FALSE)
importance_df <- importance$importance
top_10_features <- importance_df %>% arrange(desc(Overall)) %>% head(10)
print(top_10_features)## Overall
## points_price_ratio 678.84195
## is_expensive 224.44340
## price_per_point 84.02487
## points_squared 81.19805
## variety_typeRed 20.64278
## price_per_year 17.12023
## `country_regionSpain_Central Spain` 15.82920
## `country_regionUS_New York` 14.75615
## `country_regionAustralia_Australia Other` 12.65365
## country_regionFrance_Alsace 12.27927ggplot(top_10_features, aes(x = reorder(rownames(top_10_features), Overall), y = Overall)) +
geom_bar(stat = "identity", fill = "steelblue") +
coord_flip() +
labs(title = "Top 10 Important Features", x = "Features", y = "Importance (Overall)") +
theme_minimal()