dtGAP: Supervised Generalized Association Plots Based on Decision Trees

Description

Enhances decision tree visualization by incorporating Generalized Association Plots (GAP) through matrix-based visualizations including confusion matrix maps, decision tree matrix maps, and predicted class membership maps based on supervised correlation and distance metrics.

Author(s)

Maintainer: Han-Ming Wu wuhm@g.nccu.edu.tw

Authors:

• Chia-Yu Chang 110304049@g.nccu.edu.tw

• Chun-houh Chen cchen@stat.sinica.edu.tw

See Also

Useful links:

• https://github.com/hanmingwu1103/dtGAP

• Report bugs at https://github.com/hanmingwu1103/dtGAP/issues

Psychosis Disorder Data

Description

Ratings of positive and negative symptoms in psychosis disorders, based on Andreasen’s Scale for Assessment of Positive Symptoms (SAPS) and Scale for Assessment of Negative Symptoms (SANS).

Usage

Psychosis_Disorder

Format

A data frame with 95 observations and 51 variables:

UNIQID

Factor indicating disorder type.

AH1, AH2, AH3, AH4, AH5, AH6

Hallucinations subscale (SAPS).

DL1, DL2, DL3, DL4, DL5, DL6, DL7, DL8, DL9, DL10, DL11, DL12

Delusions subscale (SAPS).

BE1, BE2, BE3, BE4

Behavior subscale (SAPS).

TH1, TH2, TH3, TH4, TH5, TH6, TH7, TH8

Thought disorder subscale (SAPS).

NA1, NA2, NA3, NA4, NA5, NA6, NA7

Expression subscale (SANS).

NB1, NB2, NB3, NB4

Speech subscale (SANS).

NC1, NC2, NC3

Hygiene subscale (SANS).

ND1, ND2, ND3, ND4

Activity subscale (SANS).

NE1, NE2

Inattentiveness subscale (SANS).

Details

This data set comprises 95 subjects, of whom 69 were diagnosed with schizophrenia and 26 with bipolar disorder. All symptoms were rated on a six‐point scale (0–5).

Assigns a train/test indicator to a combined dataset

Description

Assigns a train/test indicator to a combined dataset

Usage

add_data_type(
  data_train = NULL,
  data_test = NULL,
  data_all = NULL,
  test_size = 0.3,
  seed = 42
)

Arguments

Value

A data frame with a data_type factor column.

Create Column Heatmap with Variable Importance

Description

Constructs a ComplexHeatmap object displaying feature-feature correlations with optional variable importance barplots and split-variable highlighting.

Usage

col_ht(
  fit,
  sorted_dat,
  var_imp,
  layout,
  include_var_imp = TRUE,
  col_var_imp = "orange",
  var_bar_width = 0.8,
  var_fontsize = 5,
  split_var_bg = "darkgreen",
  split_var_fontsize = 5,
  palette = "RdBu",
  n_colors = 11,
  show_col_prox = TRUE
)

Arguments

Value

A Heatmap object from ComplexHeatmap.

Compare Multiple Decision Tree Models Side-by-Side

Description

Runs the dtGAP pipeline for each specified model and composes the results side-by-side on a single wide page. Shared data preparation is performed once; each model gets its own tree + heatmap panel.

Usage

compare_dtGAP(
  models = c("rpart", "party"),
  data_train = NULL,
  data_test = NULL,
  data_all = NULL,
  target_lab = NULL,
  show = c("all", "train", "test"),
  test_size = 0.3,
  task = c("classification", "regression"),
  total_w = 594,
  total_h = 210,
  ...
)

Arguments

Value

Draws the side-by-side comparison to the current graphics device. Called for its side effect; returns invisibly.

Examples

compare_dtGAP(
  models = c("rpart", "party"),
  data_all = Psychosis_Disorder,
  target_lab = "UNIQID",
  show = "all",
  trans_type = "none",
  print_eval = FALSE
)

Compute Layout Dimensions for Tree + Heatmap Plot

Description

Determines panel widths and heights based on page dimensions, margin, and proportions.

Usage

compute_layout(
  sorted_dat,
  margin = 20,
  total_w = 297,
  total_h = 210,
  tree_p = 0.3
)

Arguments

Value

A list with:

Compute Decision Tree Data for Plotting and Analysis

Description

Builds and processes a decision tree model object to prepare data for plotting, including layout positions and terminal node summaries. need to run util.R first

Usage

compute_tree(
  fit = NULL,
  model = c("rpart", "party", "C50", "caret", "cforest"),
  show = c("all", "train", "test"),
  data = NULL,
  target_lab = NULL,
  task = c("classification", "regression"),
  custom_layout = NULL,
  panel_space = 0.001
)

Arguments

Value

A list with components:

• fit: the original fitted model

• dat: data.frame of observations with node assignments and predictions

• plot_data: data.frame of nodes with plotting variables and probabilities

• layout: data.frame of node x/y positions

Examples

library(rpart)
library(partykit)
library(ggparty)
library(dplyr)
data <- add_data_type(
  data_all = Psychosis_Disorder
)
data <- prepare_features(
  data,
  target_lab = "UNIQID",
  task = "classification"
)
fit <- train_tree(
  data = data, target_lab = "UNIQID",
  model = "rpart"
)$fit
tree_res <- compute_tree(
  fit,
  model = "rpart", show = "all",
  data = data, target_lab = "UNIQID",
  task = "classification"
)
tree_res$dat
tree_res$plot_data

Diabetes patient records.

Description

http://archive.ics.uci.edu/ml/datasets/diabetes https://www.kaggle.com/uciml/pima-indians-diabetes-database

Usage

diabetes

Format

A data frame with 768 observations and 9 variables: Pregnancies, Glucose, BloodPressure, SkinThickness, Insulin, BMI, DiabetesPedigreeFunction, Age and Outcome.

Draw Full Visualization: Decision Tree with Heatmap and Evaluation

Description

This function creates a full-page layout consisting of a decision tree plot, a heatmap, and optional evaluation results. It is designed for use in reporting classification or clustering trees with additional visual indicators.

Usage

draw_all(
  prepare_tree,
  heat,
  total_w = 297,
  total_h = 210,
  layout,
  x_eval_start = 15,
  y_eval_start = NULL,
  eval_text = 7,
  eval_res = NULL,
  print_eval = TRUE,
  show_col_prox = TRUE,
  show_row_prox = TRUE
)

Arguments

Value

Draws the full visualization to the current graphics device. Called for its side effect; returns invisible(NULL).

Examples

# See dtGAP() for a full end-to-end example
# that internally calls draw_all().

Decision Tree Generalized Association Plots (dtGAP)

Description

The dtGAP function enhances decision tree visualization by incorporating the strengths of Generalized Association Plots (GAP). While decision trees are valued for their interpretability, they often overlook deeper data structures. In contrast, GAP is effective for revealing complex associations but is typically limited to unsupervised settings. dtGAP bridges this gap by introducing matrix-based visualizations—such as the confusion matrix map, decision tree matrix map, and predicted class membership map—based on supervised correlation and distance metrics. This offers a more comprehensive and interpretable representation of decision-making processes in tree-based models.

Usage

dtGAP(
  x = NULL,
  target_lab = NULL,
  show = c("all", "train", "test"),
  model = c("rpart", "party", "C50", "caret"),
  control = NULL,
  fit = NULL,
  user_var_imp = NULL,
  data_train = NULL,
  data_test = NULL,
  data_all = NULL,
  test_size = 0.3,
  task = c("classification", "regression"),
  trans_type = c("normalize", "scale", "percentize", "none"),
  col_proximity = c("pearson", "spearman", "kendall"),
  linkage_method = c("CT", "SG", "CP"),
  seriate_method = "TSP",
  cRGAR_w = 5,
  select_vars = NULL,
  sort_by_data_type = TRUE,
  custom_layout = NULL,
  panel_space = 0.001,
  margin = 20,
  total_w = 297,
  total_h = 210,
  tree_p = 0.3,
  include_var_imp = TRUE,
  col_var_imp = "orange",
  var_imp_bar_width = 0.8,
  var_imp_fontsize = 5,
  split_var_bg = "darkgreen",
  split_var_fontsize = 5,
  Col_Prox_palette = "RdBu",
  Col_Prox_n_colors = 11,
  label_map = NULL,
  label_map_colors = NULL,
  type_palette = "Dark2",
  label_palette = "OrRd",
  n_label_color = 9,
  pred_ha_gap = unit(1, "mm"),
  prop_palette = gray,
  n_prop_colors = 11,
  Row_Prox_palette = "Spectral",
  Row_Prox_n_colors = 11,
  row_border = TRUE,
  row_gap = unit(1, "mm"),
  sorted_dat_palette = "Blues",
  sorted_dat_n_colors = 9,
  show_row_names = TRUE,
  row_names_gp = gpar(fontsize = 5),
  show_row_prox = TRUE,
  show_col_prox = TRUE,
  raw_value_col = NULL,
  lgd_direction = c("vertical", "horizontal"),
  x_eval_start = 15,
  y_eval_start = NULL,
  eval_text = 7,
  print_eval = TRUE,
  simple_metrics = FALSE,
  interactive = FALSE
)

Arguments

Value

Draws the full dtGAP visualization (decision tree + heatmap + evaluation) to the current graphics device. Called for its side effect; returns invisibly.

Examples

# Case 1: test_covid
dtGAP(
  data_train = train_covid,
  data_test = test_covid,
  target_lab = "Outcome", show = "test",
  label_map = c("0" = "Survival", "1" = "Death"),
  label_map_colors = c(
    "Survival" = "#50046d", "Death" = "#fcc47f"
  ),
  raw_value_col = colorRampPalette(
    c("#33286b", "#26828e", "#75d054", "#fae51f")
  )(9)
)
# Case 2: Psychosis_Disorder
dtGAP(
  data_all = Psychosis_Disorder,
  model = "party", show = "all",
  trans_type = "none", target_lab = "UNIQID"
)

Evaluate Tree Model Predictions and Metrics

Description

Generates summary information and confusion matrix metrics for training and/or test subsets based on a fitted decision tree and sorted matrix results.

Usage

eval_tree(
  x = NULL,
  fit = NULL,
  task = c("classification", "regression"),
  tree_res = NULL,
  target_lab = NULL,
  sorted_dat = NULL,
  show = c("all", "train", "test"),
  model = c("rpart", "party", "C50", "caret", "cforest"),
  col_proximity = c("pearson", "spearman", "kendall"),
  linkage_method = c("CT", "SG", "CP"),
  seriate_method = "TSP",
  simple_metrics = FALSE
)

Arguments

Value

A list with elements:

Examples

library(rpart)
library(partykit)
library(ggparty)
library(dplyr)
library(seriation)
data_all <- add_data_type(
  data_train = train_covid, data_test = test_covid
)
data <- prepare_features(
  data_all,
  target_lab = "Outcome",
  task = "classification"
)
train_tree <- train_tree(
  data_train = train_covid,
  target_lab = "Outcome", model = "rpart"
)
fit <- train_tree$fit
var_imp <- train_tree$var_imp
tree_res <- compute_tree(
  fit,
  model = "rpart", show = "test",
  data = data, target_lab = "Outcome",
  task = "classification"
)
sorted_dat <- sorted_mat(
  tree_res,
  target_lab = "Outcome", show = "test"
)
# Case 1: Pass the dataset name
eval_tree(
  x = "covid", fit = fit,
  task = "classification",
  tree_res = tree_res,
  target_lab = "Outcome",
  sorted_dat = sorted_dat,
  show = "test", model = "rpart"
)

Galaxy dataset for regression.

Description

Fetched from PMLB.

Usage

galaxy

Format

An object of class spec_tbl_df (inherits from tbl_df, tbl, data.frame) with 323 rows and 5 columns.

Details

#' @format A data frame with 323 observations and 5 variables: eastwest, northsouth, angle, radialposition and target (velocity).

https://www.openml.org/d/690

Generate a Bundle of Legends for Heatmap Components

Description

Creates and packs multiple legends (feature types, class labels, membership proportions, raw values, and proximity metrics) into a single legend bundle for ComplexHeatmap.

Usage

generate_legend_bundle(
  sorted_dat,
  task = c("classification", "regression"),
  show = c("all", "train", "test"),
  type_cols = NULL,
  label_cols,
  prop_cols = NULL,
  col_mat,
  col_Col_Proximity = NULL,
  col_Row_Proximity = NULL,
  direction = c("vertical", "horizontal")
)

Arguments

Value

A ComplexHeatmap packed Legend object containing all specified legends.

Build Split Factor for Heatmap Rows

Description

Extracts leaf node IDs from tree data and aligns them to the rows of the sorted proximity matrix to form a split factor.

Usage

get_split_vec(sorted_dat, tree_res)

Arguments

Value

A factor indicating leaf grouping for each row in row_pro_mat_sorted.

Draw Main Heatmap with Annotations

Description

Combines sorted data matrix and provided annotations into a single Heatmap.

Usage

make_main_heatmap(
  sorted_dat,
  split_vec,
  pred_ha,
  row_prop_ha,
  layout,
  palette = "Blues",
  n_colors = 9,
  show_row_names = TRUE,
  row_names_gp = gpar(fontsize = 5),
  show_row_prox = TRUE,
  raw_value_col = NULL
)

Arguments

Value

A configured Heatmap object.

Data of three different species of penguins.

Description

Collected and made available by Dr. Kristen Gorman and the Palmer Station, Antarctica LTER, a member of the Long Term Ecological Research Network.

Usage

penguins

Format

A data frame with 344 observations and 7 variables: species, island, culmen_length_mm, culmen_depth_mm, flipper_length_mm, body_mass_g and sex.

Gorman KB, Williams TD, Fraser WR (2014). Ecological Sexual Dimorphism and Environmental Variability within a Community of Antarctic Penguins (Genus Pygoscelis). PLoS ONE 9(3): e90081. doi:10.1371/journal.pone.0090081

Details

Fetched from https://github.com/allisonhorst/penguins.

Annotate Predsictions Information

Description

Creates row annotations showing predicted vs true class labels and probabilities, with optional data_type coloring.

Usage

prediction_annotation(
  sorted_dat,
  target_lab,
  task = c("classification", "regression"),
  label_map = NULL,
  label_map_colors = NULL,
  type_palette = "Dark2",
  label_palette = "OrRd",
  n_label_color = 9,
  prop_palette = gray,
  n_prop_colors = 11,
  gap_mm = unit(1, "mm")
)

Arguments

Value

A rowAnnotation object for predictions and truth.

Prepare Features for Modeling

Description

Converts target variable for classification tasks and coerces logical/character columns to factors.

Usage

prepare_features(
  data,
  target_lab = NULL,
  task = c("classification", "regression")
)

Arguments

Value

A tibble with processed feature types.

Prepare Tree Plot Data for Visualization

Description

This function processes a tree model's output and prepares node and segment data for visualization using ggplot2 or other plotting tools. It supports various tree model formats such as rpart, party, C50, and caret.

Usage

prepare_tree(tree_res, model = c("rpart", "party", "C50", "caret", "cforest"))

Arguments

Value

A list with two elements:

plot_data

A data frame of node-level information with labels for visualization.

branches

A data frame of edge (branch) coordinates for connecting parent and child nodes.

Examples

library(rpart)
library(partykit)
library(ggparty)
library(dplyr)
library(seriation)
data_all <- add_data_type(
  data_train = train_covid, data_test = test_covid
)
data <- prepare_features(
  data_all,
  target_lab = "Outcome",
  task = "classification"
)
train_tree <- train_tree(
  data_train = train_covid,
  target_lab = "Outcome", model = "rpart"
)
fit <- train_tree$fit
var_imp <- train_tree$var_imp
tree_res <- compute_tree(
  fit,
  model = "rpart", show = "test",
  data = data, target_lab = "Outcome",
  task = "classification"
)
prepare_tree(tree_res, model = "rpart")

Visualize a Single Tree from a Conditional Random Forest

Description

Fits a partykit::cforest and visualizes one of its individual trees using the full dtGAP pipeline (decision tree + heatmap + evaluation).

Usage

rf_dtGAP(
  x = NULL,
  target_lab = NULL,
  show = c("all", "train", "test"),
  tree_index = 1L,
  ntree = 500L,
  mtry = NULL,
  rf_control = NULL,
  data_train = NULL,
  data_test = NULL,
  data_all = NULL,
  test_size = 0.3,
  task = c("classification", "regression"),
  trans_type = c("normalize", "scale", "percentize", "none"),
  col_proximity = c("pearson", "spearman", "kendall"),
  linkage_method = c("CT", "SG", "CP"),
  seriate_method = "TSP",
  cRGAR_w = 5,
  sort_by_data_type = TRUE,
  custom_layout = NULL,
  panel_space = 0.001,
  margin = 20,
  total_w = 297,
  total_h = 210,
  tree_p = 0.3,
  include_var_imp = TRUE,
  col_var_imp = "orange",
  var_imp_bar_width = 0.8,
  var_imp_fontsize = 5,
  split_var_bg = "darkgreen",
  split_var_fontsize = 5,
  Col_Prox_palette = "RdBu",
  Col_Prox_n_colors = 11,
  label_map = NULL,
  label_map_colors = NULL,
  type_palette = "Dark2",
  label_palette = "OrRd",
  n_label_color = 9,
  pred_ha_gap = unit(1, "mm"),
  prop_palette = gray,
  n_prop_colors = 11,
  Row_Prox_palette = "Spectral",
  Row_Prox_n_colors = 11,
  row_border = TRUE,
  row_gap = unit(1, "mm"),
  sorted_dat_palette = "Blues",
  sorted_dat_n_colors = 9,
  show_row_names = TRUE,
  row_names_gp = gpar(fontsize = 5),
  show_row_prox = TRUE,
  show_col_prox = TRUE,
  raw_value_col = NULL,
  lgd_direction = c("vertical", "horizontal"),
  x_eval_start = 15,
  y_eval_start = NULL,
  eval_text = 7,
  print_eval = TRUE,
  simple_metrics = FALSE
)

Arguments

Value

Draws the dtGAP visualization for the selected tree to the current graphics device. Called for its side effect; returns invisibly.

Examples

rf_dtGAP(
  data_train = train_covid,
  data_test = test_covid,
  target_lab = "Outcome",
  show = "test",
  tree_index = 1,
  ntree = 50,
  print_eval = FALSE
)

Random Forest Ensemble Summary

Description

Fits a partykit::cforest and displays a multi-panel summary: variable importance barplot, OOB error curve, and optionally a representative tree (the tree with highest prediction agreement with the full ensemble).

Usage

rf_summary(
  x = NULL,
  target_lab = NULL,
  data_train = NULL,
  data_test = NULL,
  data_all = NULL,
  test_size = 0.3,
  task = c("classification", "regression"),
  ntree = 500L,
  mtry = NULL,
  rf_control = NULL,
  show_var_imp = TRUE,
  show_rep_tree = TRUE,
  top_n_vars = 15L,
  total_w = 297,
  total_h = 210
)

Arguments

Value

A list (invisible) with:

Examples

rf_summary(
  data_train = train_covid,
  data_test = test_covid,
  target_lab = "Outcome",
  ntree = 50
)

Annotate Row Proximity on Heatmap

Description

Creates a ComplexHeatmap row annotation showing supervised proximity for each data sample, grouped by a split vector.

Usage

row_prop_anno(
  sorted_dat,
  layout,
  split_vec,
  palette = "Spectral",
  n_colors = 11,
  border = TRUE,
  gap_mm = unit(1, "mm"),
  show_row_prox = TRUE
)

Arguments

Value

A rowAnnotation object for use in a ComplexHeatmap.

Save dtGAP Visualization to File

Description

Exports the dtGAP plot to PNG, PDF, or SVG format.

Usage

save_dtGAP(
  file,
  format = NULL,
  width = 297,
  height = 210,
  dpi = 300,
  bg = "white",
  ...
)

Arguments

Value

Invisible file path of the created file.

Examples

save_dtGAP(
  file = tempfile(fileext = ".png"),
  data_train = train_covid,
  data_test = test_covid,
  target_lab = "Outcome",
  show = "test",
  print_eval = FALSE
)

Performs transformation on continuous variables.

Description

Performs transformation on continuous variables for the heatmap color scales.

Usage

scale_norm(x, trans_type = c("percentize", "normalize", "scale", "none"))

Arguments

Value

Numeric vector of the transformed x.

References

https://github.com/trangdata/treeheatr/blob/85be4a61e35a62285c95b553f03729721bb18a0b/R/utils.R

Examples

scale_norm(1:5, "normalize")

Sort Feature Matrix by Tree and Correlation Structure

Description

Orders samples and features based on tree-derived node grouping and correlation-based seriation.

Usage

sorted_mat(
  tree_res = NULL,
  target_lab = NULL,
  show = c("all", "train", "test"),
  trans_type = c("normalize", "scale", "percentize", "none"),
  col_proximity = c("pearson", "spearman", "kendall"),
  linkage_method = c("CT", "SG", "CP"),
  seriate_method = "TSP",
  w = 5,
  sort_by_data_type = TRUE
)

Arguments

Value

A list with:

• sorted_row_names, sorted_col_names

• row_pro_mat_sorted, col_pro_mat_sorted

• cRGAR_score

• sorted_test_matrix

• node_ids

• dat_sorted

Examples

library(rpart)
library(partykit)
library(ggparty)
library(dplyr)
library(seriation)
data <- add_data_type(
  data_all = Psychosis_Disorder
)
data <- prepare_features(
  data,
  target_lab = "UNIQID",
  task = "classification"
)
fit <- train_tree(
  data = data, target_lab = "UNIQID",
  model = "rpart"
)$fit
tree_res <- compute_tree(
  fit,
  model = "rpart", show = "all",
  data = data, target_lab = "UNIQID",
  task = "classification"
)
sorted_dat <- sorted_mat(
  tree_res,
  target_lab = "UNIQID",
  show = "all", trans_type = "none",
  seriate_method = "GW_average",
  sort_by_data_type = FALSE
)
sorted_dat$row_pro_mat_sorted
sorted_dat$col_pro_mat_sorted
sorted_dat$cRGAR_score

External test dataset. Medical information of Wuhan patients collected between 2020-01-10 and 2020-02-18.

Description

External test dataset. Medical information of Wuhan patients collected between 2020-01-10 and 2020-02-18.

Usage

test_covid

Format

A data frame with 110 observations and 7 XGBoost-selected variables: PATIENT_ID, Lactate dehydrogenase, High sensitivity C-reactive protein, (%)lymphocyte, Admission time, Discharge time and outcome.

An interpretable mortality prediction model for COVID-19 patients. Yan et al. https://doi.org/10.1038/s42256-020-0180-7 https://github.com/HAIRLAB/Pre_Surv_COVID_19

Training dataset. Medical information of Wuhan patients collected between 2020-01-10 and 2020-02-18. Containing NAs.

Description

Training dataset. Medical information of Wuhan patients collected between 2020-01-10 and 2020-02-18. Containing NAs.

Usage

train_covid

Format

A data frame with 375 observations and 77 variables.

An interpretable mortality prediction model for COVID-19 patients. Yan et al. https://doi.org/10.1038/s42256-020-0180-7 https://github.com/HAIRLAB/Pre_Surv_COVID_19

Fit a Conditional Random Forest

Description

Fits a conditional random forest using partykit::cforest() and returns the forest object along with variable importance scores.

Usage

train_rf(
  data_train,
  target_lab,
  task = c("classification", "regression"),
  ntree = 500L,
  mtry = NULL,
  control = NULL
)

Arguments

Value

A list with elements:

Examples

data(train_covid)
rf_res <- train_rf(train_covid, target_lab = "Outcome", ntree = 50)
rf_res$var_imp

Fit a Decision Tree Model

Description

Fits a decision tree to training data using one of several supported tree implementations (rpart, party, C50, or via caret) and returns a standardized party object along with variable importance scores.

Usage

train_tree(
  data_train = NULL,
  data = NULL,
  target_lab = NULL,
  model = c("rpart", "party", "C50", "caret"),
  task = c("classification", "regression"),
  control = NULL
)

Arguments

Value

A list with elements:

Examples

library(partykit)
library(C50)
library(caret)

data(train_covid)
train_tree(data_train = train_covid, target_lab = "Outcome", model = "rpart")
train_tree(data_train = train_covid, target_lab = "Outcome", model = "C50")
train_tree(data_train = train_covid, target_lab = "Outcome", model = "caret")

data(Psychosis_Disorder)
data <- add_data_type(data_all = Psychosis_Disorder)
data <- prepare_features(data, target_lab = "UNIQID", task = "classification")
train_tree(
  data = data, target_lab = "UNIQID", model = "party",
  control = ctree_control(minbucket = 15)
)

Results of a chemical analysis of wines grown in a specific area of Italy.

Description

Three types of wine are represented in the 178 samples, with the results of 13 chemical analyses recorded for each sample.

Usage

wine

Format

A data frame with 178 observations and 14 variables: Alcohol, Malic, Ash, Alcalinity, Magnesium, Phenols, Flavanoids, Nonflavanoids, Proanthocyanins, Color, Hue, Dilution, Proline and Type (target).

Details

Import with data(wine, package = 'rattle'). Dependent variable: Type. https://rdrr.io/cran/rattle.data/man/wine.html http://archive.ics.uci.edu/ml/datasets/wine

Red variant of the Portuguese "Vinho Verde" wine.

Description

Fetched from PMLB. Physicochemical and quality of wine.

Usage

wine_quality_red

Format

A data frame with 1599 observations and 12 variables: fixed.acidity, volatile.acidity, citric.acid, residual.sugar, chlorides, free.sulfur.dioxide, total.sulfur.dioxide, density, pH, sulphates, alcohol and target (quality).

http://archive.ics.uci.edu/ml/datasets/Wine+Quality

P. Cortez, A. Cerdeira, F. Almeida, T. Matos and J. Reis. Modeling wine preferences by data mining from physicochemical properties. In Decision Support Systems, Elsevier, 47(4):547-553, 2009.