IDE IAT 2 Flashcards

Logistic regression models the probability of a categorical outcome using the __________ function.

logistic (sigmoid)

The natural logarithm of the odds is called the __________ of the odds.

logit

The ratio of the probability that an event occurs to the probability that it does not occur is the __________.

odds

Maximum likelihood estimation chooses parameter values that make the observed data most __________.

likely

In binary logistic regression, the response variable has __________ possible categories.

two

The likelihood ratio test compares a fitted model with a __________ model.

null (reduced)

Pseudo R-squared measures provide an approximate indication of model __________; they are not identical to

ordinary least-squares R-squared.

fit (goodness of fit)

The Hosmer–Lemeshow test assesses how well predicted probabilities agree with __________ outcomes.

observed

The Wald test evaluates whether an estimated model __________ differs significantly from a specified value.

coefficient (parameter)

When selecting a model for prediction, both predictive performance and model __________ should be

considered.

interpretability (complexity)

A decision tree predicts a class by following a sequence of __________ from the root to a leaf.

decisions (splits)

In a classification tree, terminal nodes assign a predicted __________.

class

The depth of a tree describes its __________ structure.

hierarchial

Data used to fit a decision tree are called the __________ set.

training

Data held back to evaluate performance are called the __________ set.

test

An induction algorithm selects splits using a criterion such as information gain or __________ impurity.

Gini

Probability estimates in a decision tree can be based on class __________ within a leaf.

proportions

Tree growth may be stopped when a node contains too few observations or when further splits fail a stopping

__________.

criterion (rule)

A confusion matrix compares actual classes with __________ classes.

predicted

A ROC curve plots the true positive rate against the __________ positive rate.

false

A lift curve compares a model’s performance with a __________ selection strategy.

random

McNemar’s test can compare the error patterns of two classifiers on the same __________ cases.

test

In k-fold cross-validation, the data are divided into __________ roughly equal parts.

k

A paired t-test evaluates the mean difference between paired model-performance __________.

scores / metrics / measures

Combining multiple trees is known as a tree __________ method.

ensemble

Kernel methods can represent nonlinear relationships through an implicit mapping into a feature __________.

space

SVM stands for support vector __________.

machine

SVR stands for support vector __________.

regression

The SVM decision boundary is chosen to maximize the __________ between classes.

margin

The separating boundary in an SVM is called a __________ hyperplane.

optimal separating / maximum-margin

In the primal formulation, the model is expressed using the original optimization __________.

variables

The dual formulation expresses the optimization in terms of __________ variables.

Lagrange multiplier

A soft-margin SVM allows some margin violations, controlled by a regularization __________.

parameter

A kernel function computes similarities as if observations had been mapped into a higher-dimensional __________ space.

feature

The linear, polynomial, radial basis, and sigmoid are examples of common __________ functions.

kernel

The radial basis function kernel is often abbreviated __________.

RBF

Unsupervised learning finds patterns in data without using labeled __________ values.

target

Clustering groups observations according to their __________ or dissimilarity.

similarity

Principal component analysis (PCA) transforms data into orthogonal components that capture as much __________ as possible.

variance