CAP Predictive Analytics

Question 1

Predictive analytics

Answer 1

Statistical techniques used to make predictions about future or otherwise unknown events

Question 2

Data science

Answer 2

Field that applies statistics, data analysis, machine learning, and data mining in order to understand and analyze phenomena

Question 3

Big Data

Answer 3

Data sets so large or complex that traditional data processing applications are inadequate

Question 4

Machine learning

Answer 4

Study of algorithms and models that learn to perform a specific task based on pattern recognition

Question 5

Data mining

Answer 5

The process of discovering patterns in large data sets

Question 6

Deep learning

Answer 6

A family of machine learning methods based on learning data representations as opposed to task-specific algorithms

Question 7

Linear regression

Answer 7

Approach to modeling the relationship between a dependent variable and one or more independent variables in a linear fashion

Question 8

Support Vector Machine

Answer 8

A classifier that attempts to find the maximum separating hyperplane between samples in two classes

Question 9

K-nearest neighbor (K-NN)

Answer 9

A classifier that assigns new samples the majority classification of the k nearest samples in the training space

Question 10

Logistic regression

Answer 10

Approach to modeling the relationship between continuous independent variables and a binary dependent variable, uses a sigmoid function

Question 11

Decision tree

Answer 11

A classifier that uses root nodes, branches, and leaf nodes to indicate classification decisions on unknown samples; learning comes from maximizing information gain or other metrics at each node

Question 12

k-means clustering

Answer 12

Unsupervised learning method that groups elements into one of k different clusters based on which cluster it is closest to

Question 13

Random forest

Answer 13

Ensemble method that randomly selects features, trains a set of decision trees using those features, and then uses majority voting for classification of samples

Question 14

Naive Bayes Classifier

Answer 14

Probabilistic classifier that applies Bayes’ theorem with strong independence assumptions about features

Question 15

Neural network

Answer 15

A classifier that uses artificial neurons/perceptrons linked to each other to learn patterns in data

Question 16

AdaBoost

Answer 16

Boosting algorithm that trains a series of weak learners on data and focuses later learners on data misclassified by earlier learners

Question 17

Kernel factory

Answer 17

An ensemble method that uses kernel machines as classifiers and a genetic algorithm to tune the weights of each machine’s vote in classification

Question 18

Rotation Forest

Answer 18

An ensemble method that uses decision trees as classifiers and principal components of randomly selected variables from training data set as features

Question 19

Dimension reduction

Answer 19

Methods that reduce the number of features used as inputs to a classifier or regressor

Question 20

Gradient boosting

Answer 20

a machine learning technique for regression and classification problems, which produces a prediction model in the form of an ensemble of weak prediction models, typically decision trees. It builds the model in a stage-wise fashion and it generalizes them by allowing optimization of an arbitrary differentiable loss function

Question 21

Ensemble

Answer 21

Machine learning method that aggregates multiple models into one model

Question 22

Bagging; bootstrap aggregating

Answer 22

Ensemble method that uses sampling with replacement to build training data sets for multiple machine learning methods; intended to decrease variance

Question 23

Boosting

Answer 23

Ensemble method that uses weighted sampling based on errors made by previous models to build training data sets for future models; intended to decrease bias

Question 24

Stacking

Answer 24

Ensemble method that trains multiple different types of models on full training data set, then uses those predictions as features to train a meta-method; intended to increase predictive performance