What is a 'neural network'?

Question 1 options:

A regression-based exponential smoothing model.

An algorithm that is based on the interconnected neurons seen in biology.

A static model based only on internal (autoregressive) information.

A group of experts who work together to forecast.

Question 2

According to the text, evidence indicates that neural networks are most beneficial in forecasts involving

Question 2 options:

Low frequency series (annual data)

Very short series (few observations)

High frequency series (hourly or daily data for example)

Qualitative data series

Question 3

The 3 types of layers in a neural network are

Question 3 options:

Training, Validation and Future

Huey, Duey and Louie

Input, Hidden and Output

Daily, Weekly and Quarterly

Question 4

The nodes in a neural network diagram represent

Question 4 options:

Variables

Calculations

Decisions

Assumptions

Question 5

The nodes in the hidden layer are

Question 5 options:

Unobservable or unknown to the user

The weighted sum of the inputs to which an activation function has been applied

Assigned by the forecaster

Chosen at random

Question 6

A simple neural network that contains no hidden nodes is

Question 6 options:

The Delphi method

The same as a nave forecast

Comparable toa MA(1) model

Equivalent to linear regression

Question 7

In a neural network autoregression, to scale the data so that it is in the interval [-1, 1] you can

Question 7 options:

Take first differences and divide by the absolute value

Divide each observation by the maximum of the absolute values of the series

Subtract the absolute value from each observation

Any of these strategies are acceptable

Question 8

The 'forecast' package in R includes the function nnetar which

Question 8 options:

Determines whether a neural network model has superior fit compared to an AR model

Returns the values of the hidden nodes in a neural network

Can be used for neural network autoregression with seasonal or non-seasonal data

Is identical to the avNNet function in the 'caret' package

Question 9

How should one determine the number of hidden layers and the number of node in each hidden layer when creating a neural network?