08_Machine_Learning

Question 1

Front

Answer 1

Back

Question 2

How do you fit a decision tree in R with rpart?

Answer 2

Use rpart() with method = ‘class’ (classification) or ‘anova’ (regression).

Code:
library(rpart); library(rpart.plot)
fit <- rpart(Species ~ ., data = iris, method=’class’)
rpart.plot(fit)
predict(fit, newdata = iris, type=’class’)

Question 3

How do you run k-means clustering?

Answer 3

Use kmeans(x, centers, nstart). Scale numeric features first.

Code:
x <- scale(iris[,1:4])
set.seed(1)
km <- kmeans(x, centers = 3, nstart = 25)
table(km$cluster, iris$Species)

Question 4

How do you train a model with caret::train()?

Answer 4

Unified interface for resampling and tuning.

Code:
library(caret)
ctrl <- trainControl(method=’cv’, number=5)
fit <- train(Species ~ ., data=iris, method=’rf’, trControl=ctrl)
confusionMatrix(fit)

Notes:
method names: ‘rf’, ‘rpart’, ‘glmnet’, etc. Use tuneGrid for hyperparameters.

Question 5

How do you evaluate classification predictions?

Answer 5

Use confusionMatrix() and yardstick metrics.

Code:
library(caret)
pred <- predict(fit, newdata=iris)
confusionMatrix(pred, iris$Species)