MLR3

Question 1

It is the interface to (almost) all implementations of machine learning algorithms in R, it is unified and hence user-friendly access to > 100 supervised learning techniques.

Answer 1

MLR3

Question 2

They wrap a DataBackend, and store meta-information, such as the role of the individual columns in the DataBackend.

Answer 2

MLR3 Tasks

Question 3

An object to transparently interface different data storage types.

Answer 3

Data Backend

Question 4

The conceptual description of the machine learning algorithm. An “abstract hull” of the model.

Answer 4

Learner

Question 5

Which R function should we use to get a task from the mlr3 task dictionary?

Answer 5

tsk ()

Question 6

What is the notation to create a new classification task in R?

Answer 6

task <- TaskClassif$new()

Question 7

What is the notation to add a created task to the task dictionary in R?

Answer 7

mlr_tasks$add ()

Question 8

Concept of mlr3 that consists in training the supervised learning models based on the task.

Answer 8

“train”

Question 9

Concept of mlr3 that consists on using the models to make predictions (on new data)

Answer 9

“predict”

Question 10

Two concepts of mlr3 that assess a model’s quality.

Answer 10

Performance & resampling

Question 11

Comparison of ≥ 1 learner(s) across ≥ 1 task(s)

Answer 11

Benchmarking

Question 12

Optimizing the learner’s hyperparameters

Answer 12

Tuning

Question 13

Looking for a ‘better’ subset of features

Answer 13

Feature selection

Question 14

(True or false) When creating a task, the target variable has to be specified.

Answer 14

True

Question 15

Five most popular machine learning methods for mlr3

Answer 15

*linear and logistic regression
*k-nearest neighbor methods
* support vector machines
* gradient boosting
* random forests

Question 16

This function randomly splits the task into two disjoint sets: a training set (67% of the total data, the default) and a test set (33% of the total data, the data not part of the training set).

Answer 16

partition ()

Question 17

This function restricts tasks to a given set of features.

Answer 17

$select()

Question 18

This function applies restriction to rows (observations)

Answer 18

$filter() with the row IDs

Question 19

Which are the two roles a column of a dataset can have in mlr3?

Answer 19

Targets and features

Question 20

Which information (6) is stored in a learner?

Answer 20

• Model
• Parameters
• Packages
• Predict Type
• Feature type
• Properties

Question 21

With which function can you access a learner’s hyperparameters?

Answer 21

$param_set

Question 22

(True or false) A model allows a user to set their own hyperparameters.

Answer 22

False. A model automatically learns its own (internal) parameters such as the splits of a decision tree or a neural network’s weights.

Question 23

(True or false) The learner’s $model element
is always empty before training.

Answer 23

True

Question 24

(True or false) At the predict stage, all parameters and hyperparameters of the model are fixed.

Answer 24

True

Question 25

What are the two types of prediction?

Answer 25

Directly on the task or on a new dataset

Question 26

Loss as measured when using the same data set for training and assessing a model

Answer 26

Apparent error (or resubstitution error)

Question 27

This type of resampling splits data into a training set and a test set.

Answer 27

Holdout approach

Question 28

Helps to assess model quality when data set is too small for simple holdout approach.

Answer 28

Resampling

Question 29

Split data randomly into 𝑘 blocks of (roughly) equal size, then create 𝑘 training-test-splits (called folds) and each of the 𝑘 blocks acts as test data exactly once

Answer 29

k Fold cross validation

Question 30

Variant of cross validation in which every observation is a block itself.

Answer 30

Leave-one-Out (LOO) Crossvalidation

Question 31

Variant of cross validation that repeats 𝑘-fold CV multiple times (with different data-to-fold assignments for the different replications).

Answer 31

Repeated Cross Validation

Question 32

Create 𝐵 samples of training-test-splits and each of the 𝐵 splits is created by drawing randomly with replacement 𝑛 observations from the original data set (of size 𝑛)

Answer 32

Bootstrap Sampling

Question 33

Combines concepts of bootstrap and holdout: creates 𝐵 samples of training-test-splits and each of the 𝐵 splits is created by drawing randomly without replacement a predefined fraction from the original data set.

Answer 33

Subsampling

Question 34

Function used when one wants to compare multiple learners across one (or more) task(s).

Answer 34

Benchmarking

Question 35

True or false? Leave-one-out crossvalidation is equal to n-fold crossvalidation if n is the size of the data set.

Answer 35

True

Question 36

True or false? The models which are trained during resampling should be used to make predictions.

Answer 36

False. Resampling is a method to estimate the generalization error of a model. To make predictions, another model should be trained using the whole data set.

Question 37

True or false? The resampling method is a tunable model parameter.

Answer 37

False. The resampling method is not a parameter of the machine learning model.