Partitioning Trees

Author

Affiliation

Martin Bari Garnier

Published

January 24, 2024

Code

load("../data/train.dtf.Rdata")
load("../data/test.dtf.Rdata")

Code

library(rpart)
library(rpart.plot)

Création de jeux de données corrects

Code

#train.names <- sample(rownames(X), 2*nrow(X)/3, replace = FALSE)
#test.names <- rownames(X)[-which(rownames(X) %in% train.names)]

#res.sample <- PCA(X, scale.unit = TRUE, ind.sup = match(test.names, rownames(X)))
#fviz_pca_biplot(res.sample, geom='point')

#train.set <- X[train.names,]
#test.set <- X[test.names,]

#train.dtf <- cbind(train.set, Y[train.names,])
#test.dtf <- cbind(test.set, Y[test.names,])

#save(train.dtf, file = "../data/train.dtf.Rdata")
#save(test.dtf, file = "../data/test.dtf.Rdata")

Code

train_drugg <- train.dtf[, -which(colnames(train.dtf) == "score")]
test_drugg <- test.dtf[, -which(colnames(test.dtf) == "score")]

Fit <- rpart(drugg~., data = train_drugg, control = rpart.control(minsplit = 1))
rpart.plot(Fit, extra = 2)

Code

library(caret)

Loading required package: ggplot2

Loading required package: lattice

Code

# Train set
fit_predict_train <- predict(Fit, newdata = train_drugg)
conf_train_fit <- confusionMatrix(as.factor(round(fit_predict_train[, 2])),
                               as.factor(train_drugg$drugg))
# Test set
fit_predict_test <- predict(Fit, newdata = test_drugg)
conf_test_fit <- confusionMatrix(as.factor(round(fit_predict_test[, 2])),
                               as.factor(test_drugg$drugg))
# Plot
plotcp(Fit)

Code

Ness <- prune(Fit, cp = 0.057)
rpart.plot(Ness)

Code

# Train set
ness_predict_train <- predict(Ness, newdata = train_drugg)
conf_train_ness <- confusionMatrix(as.factor(round(ness_predict_train[, 2])),
                               as.factor(train_drugg$drugg))

# Test set
ness_predict_test <- predict(Ness, newdata = test_drugg)
conf_test_ness <- confusionMatrix(as.factor(round(ness_predict_test[, 2])),
                               as.factor(test_drugg$drugg))

Code

conf_train_fit$overall[1]

Accuracy 
       1

Code

conf_test_fit$overall[1]

 Accuracy 
0.7027027

Code

conf_train_ness$overall[1]

 Accuracy 
0.9305556

Code

conf_test_ness$overall[1]

 Accuracy 
0.7027027

Code

df_acc <- data.frame(
  train_fit = conf_train_fit$overall[1],
  test_fit = conf_test_fit$overall[1],
  train_ness = conf_train_ness$overall[1],
  test_ness = conf_test_ness$overall[1]
)

df_acc <- t(df_acc)
df_acc

            Accuracy
train_fit  1.0000000
test_fit   0.7027027
train_ness 0.9305556
test_ness  0.7027027

Code

train_score <- train.dtf[, -which(colnames(train.dtf) == "drugg")]
test_score <- test.dtf[, -which(colnames(test.dtf) == "drugg")]

Fit <- rpart(score~., data = train_score, control = rpart.control(minsplit = 2))
rpart.plot(Fit)

Code

library(caret)

# Train set
fit_spred_train <- predict(Fit, newdata = train_score)
sperf_train_fit <- postResample(fit_spred_train, train_score$score)

# Test set
fit_spred_test <- predict(Fit, newdata = test_score)
sperf_test_fit <- postResample(fit_spred_test, test_score$score)

# Plot
plotcp(Fit)

Code

Ness <- prune(Fit, cp = 0.042)
rpart.plot(Ness)

Code

# Train set
ness_spred_train <- predict(Ness, newdata = train_drugg)
sperf_train_ness <- postResample(ness_spred_train, train_score$score)

# Test set
ness_spred_test <- predict(Ness, newdata = test_drugg)
sperf_test_ness <- postResample(ness_spred_test, test_score$score)

Code

# Create a dataframe
results_df <- data.frame(
  RMSE = c(41.3386484, 50.6410931, 46.5379404, 50.6410931),
  Rsquared = c(0.9229773, 0.9299857, 0.9023842, 0.9299857),
  MAE = c(33.3168518, 39.2805625, 37.7836200, 39.2805625),
  row.names = c("train_original", "test_original", "train_pruned", "test_pruned")
)

Code

print("hello world!")

Reuse

CC-BY-SA-4.0

Citation

BibTeX citation:

@online{bari garnier2024,
  author = {Bari Garnier, Martin},
  title = {Partitioning {Trees}},
  date = {2024-01-24},
  url = {https://MartinBaGar.github.io/Master_ISDD_fiches//mda/pages/tp4.html},
  langid = {en}
}

For attribution, please cite this work as:

M. Bari Garnier, Partitioning Trees, (2024). https://MartinBaGar.github.io/Master_ISDD_fiches//mda/pages/tp4.html.