AN 420 Final Exam UKY

What is the algebraic formula used to create predictions in a linear regression model?

y = mx + b

True or false: Value ranges for all attributes for every observation in a scoring data set must be within the value ranges for the corresponding attributes in the training data set in a linear regression model.

True

In linear regression, the sum of the confidence level and alpha (α) is always ________.

1

In linear regression, the p-values for each independent variable must be smaller than ________.

alpha

In linear regression, the m variable is the independent variable's ________.

Coefficient

In linear regression, the x variable is the independent variable's ________.

value

In linear regression, the b variable is the model's ________.

Intercept coefficient

In linear regression, p-values larger than alpha indicate that their corresponding independent variables are __________.

Not statistically significant

The data type of the dependent variable in linear regression must be ________.

Continuous numeric

The data types of all independent variables in linear regression must be _________.

Numeric

In linear regression, the R-squared value indicates the __________ between the dependent variable and the independent variable(s).

Percent of shared variability

T/F: If you attempt to make a prediction for an out-of-range scoring observation in a linear regression model in RapidMiner, the software will throw an error.

false

True or false: In logistic regression, the smaller the p-Value for an independent variable, the more predictive power that variable has relative to the dependent variable.

True

The data type of the dependent variable in logistic regression must be ________.

Binary

A date or date/time

Continuously numeric

A string

Binary

The data types of all independent variables in logistic regression must be _________.

Numeric

In a logistic regression model, the confidence values that indicate how sure you can be that the binary prediction is correct are called _________ percentages.

Post-probability

True or false: The values true/false or 0/1 would both be valid combinations for the dependent variable in a logistic regression model.

True

True or false: Unlike in linear regression, it is possible to have more than one dependent variable in a logistic regression model.

False

In a logistic regression model, if all p-Values are rounded to zero, you can determine the relative predictive power of independent variables using the _________.

z-Value

The default confidence percent used for logistic regression models is _______.

95%

In logistic regression models, if the predicted confidence percent is 50% or greater, the class prediction will be __________.

True

False

0

Neither True nor False

True

True or false: There is more than one algorithm available for use when producing logistic regression models.

True

In RapidMiner, the data type of the dependent variable in a logistic regression model must be __________.

Binominal

In RapidMiner, a logistic regression model will produce _________ when applied to a scoring data set using the Apply Model operator.

A binary class prediction and confidence percentages for the positive and negative dependent variable outcomes

True or false: In RapidMiner, the label (dependent variable) can be coded either alphabetically (e.g., true/false) or numerically (e.g., 0/1).

True

True or false: The Logistic Regression operator in RapidMiner offers only one algorithm for model creation.

False

If a training data set in RapidMiner contains a non-predictive, numeric identification column, how must this be handled when creating logistic regression models?

The role for the identification column must be set to "ID."

In logistic regression in R, the glm command must include a parameter setting the family equal to ___________.

binominal

In a decision tree, the independent variable found at each branch of the tree is known as a _________.

Node

True or false: In decision tree models, no independent variable can be used more than once.

false

True or false: In a decision tree model, not all training observations that follow a specific path through the tree must have the same dependent variable outcome.

true

Data types for independent variables in a decision tree model must be ___________.

Numeric

Binary

Text

Any of the above

Any of the above

If a data analyst finds that a decision tree model has too many nodes or leaves to be meaningful, the analyst should apply _________ to the tree.

pruning

True or false: Unlike some other predictive modeling techniques, decision tree models do not provide confidence percentages alongside their predictions.

false

In decision trees, CART is an acronym that stands for ________.

Classification and Regression Trees

True or false: In decision tree models, all independent variables are given equal weight when making predictions.

False

The data type for the dependent variable in a classification decision tree model must be __________.

Nominal

In a decision tree, the dependent variable value found at the end of each path through the tree is known as a _________.

Leaf

In a decision tree model represented visually in RapidMiner, the first predictive independent variable is represented __________.

At the top

Increasing which parameter of the Decision Tree operator in RapidMiner would reduce the size of the tree?

Minimal Leaf Size

How many algorithm options for constructing tree models are there in the Decision Tree operator in RapidMiner?

5

Neural networks build probability pathways between combinations of independent variable values and dependent variable outcomes through a process called forward and back _________.

Propagation

The data type required for independent variables in a neural network model must be _________.

Numeric

The space between the independent variables and the dependent variable where a neural network model gets trained is called the ____________.

Hidden layer

Inferential, probability-based approaches to data comparisons allowing inference based on probabilities to determine the strength of the relationship between attributes in data sets is known as __________.

Fuzzy logic

When training a neural network, the process of mapping independent variables forward to the dependent variable and then backward to the independent variables will repeat until __________ is reached.

Convergence

In neural networks, the pathways between independent variables and dependent variables are called __________.

Synapses

What is calculated in the nodes of the hidden layer of a neural network?

Independent variable weights

In a neural network model, how many nodes will the output layer always have?

The number of distinct values in the dependent variable

Which formula represents the general guideline for choosing the size of a neural network's hidden layer?

(((count(independent variables) + count(dependent variable levels))/2)+1

What do we call the process of tracing a neural network pathway from a dependent variable outcome to its independent variable inputs?

Backpropagation

In RapidMiner, if the number of hidden layers is not specified by the analyst, how many hidden layers will be used to train a neural network?

1

In RapidMiner, which of the following will automatically be generated when the Apply Model operator applies a neural network model to a scoring data set?

Both class predictions and confidence percentages

In the Neural Net operator in RapidMiner, which of the following parameters will cause the model to stop the training process if its value is reached?

Training cycles

Momentum

Learning rate

All of these would stop training if their value is reached.

None of these would stop training if their value is reached.

All of these would stop training if their value is reached.

In RapidMiner, if one or more independent variables has a non-numeric data type, what would be required for the Neural Net operator to work correctly?

The non-numeric independent variables could be recoded to numeric values or excluded from the model.

In R, the type parameter required to see category predictions from a neural network using the predict function is __________.

class

The process of converting words from text into data points in text mining is called _________.

Tokenization

Words required for sentences to make grammatical sense in written language, but that are not helpful in text mining results are called __________.

Stopwords

Phrases consisting of two or more words that are combined together in text mining results in order to retain context are called __________.

n-grams

Combining similar words such as "nation," "nations," "national," and "nationality" into a single data point in text mining is called _________.

Stemming

To ensure that text mined terms such as "Complaint" and "complaint" are identified as the same data point, an analyst should __________.

Transform cases

True or false: In text mining, if the data analyst wants data elements that are similar (e.g., car, truck, van = vehicle), it is possible to replace these data items with a single representative item

True

True or false: Because text mining deals with words in paragraphs, it is not possible to create charts or graphs to visualize results.

TrueFalse

False

True or false: Data analysts can create their own lists of words to be removed during text mining activities.

True

The text from one or more documents that is analyzed during text mining is referred to as the _________.

Corpus

Which category of data mining and analytics is most descriptive of text mining activities?

Unstructured data analysis

The process of checking for the likelihood of false positives in predictive models is called _________.

Cross-validation

True or false: A false positive is when a model predicts an expected outcome incorrectly.

True

In order to test a predictive model's accuracy, apply the model to the _________ data, then compare predicted values to the dependent variable.

training

True or false: Cross-validation is not necessary if a model produces usable predictions without it.

False

True or false: It is not possible to validate a classification model such as k-Means.

False

True or false: In addition to accuracy rates, cross-validation can also provide a data analyst with error rates.

True

The "k" in k-folds indicates ___________.

The number of groups the training data is segmented into

True or false: When using k-folds cross-validation, k should be set to 10.

False

Cross-validation can determine the predictive accuracy of all of the following except __________.

k-Means clustering

Cross-validation can determine predicted categorical outcomes for all of the following except __________.

Linear regression

True or false: In RapidMiner, the performance operator you choose is dictated by the type of modeling technique you are validating.

True

True or false: When cross-validating a model in RapidMiner, the appropriate Performance operator to be used in the subprocess will depend on the type of dependent variable in the training data.

True

A Performance (Classification) operator in RapidMiner will automatically generate which of the following validation outputs?

A model's predictive accuracy

Requiring user authentication before allowing access to digital data is an example of which of Lawrence Lessig's mechanisms for governing ethical behavior?

code

The set of moral codes above and beyond the legally required minimums that an individual uses to make right and respectful decisions is called _________.

ethics

Publicly disclosing the kinds and extents of data collected by a mobile app and allowing people to have input on the use of such is an example of which of Lawrence Lessig's mechanisms for governing ethical behavior?

social norms

People refusing to use a given social media platform because of the sale of user data for data mining is an example of which of Lawrence Lessig's mechanisms for governing ethical behavior?

markets

Governmental requirements to report unauthorized access to hospital patient information is an example of which of Lawrence Lessig's mechanisms for governing ethical behavior?

laws

Which ethical framework is defined in the following quote?

"Unless a person can take a given action repeatedly without causing harm, that person should not take that action even once."

Descartes' rule of change

Kant's categorical imperative

Thoreau's value expectation

Voltaire's moral standard

Descartes' rule of change

Which ethical framework is defined in the following quote?

"Unless all members of a society can take a given action without causing harm, then no members of that society should take that action."

Thoreau's value expectation

Descartes' rule of change

Kant's categorical imperative

Voltaire's moral standard

Kant's categorical imperative

Which of the following is a professional organization that provides a code of ethics that can be used by data miners and analysts?

Association for Computing Machinery

Data mining ethics includes respect for __________.

Privacy

Accuracy

Confidentiality

All of the above

All of the above

True or false: Because privacy and confidentiality are so important, it is in everyone's best interest to collect and analyze data quietly in the background of an organization.

False

Which of the following is NOT an organization that maintains a professional code of ethics that is relevant to data analysts?

ANA

Linear Regression

predictive data mining method that uses the algebraic formula for calculating the slope of a line to predict where a given observation will likely fall during that line

statistically significant

the measure of whether or not the model has yielded any results that are mathematically reliable enough to be used

confidence interval

the probability that an estimated value in an analytic model, created using a data sample, is also true for the population represents in the sample

alpha

the probability of rejecting a null hypothesis. Alpha is usually 5% leaving CL = 95%

logistic regression

a predictive data mining method that uses a quadratic formula to predict one of a set of possible outcomes, along with a probability that the prediction will be the actual outcome

neural network

a predictive data mining methodology that tries to mimic human brain process by comparing that values of all attributes in a date set to one another through the use of a hidden layer of nodes.

fuzzy logic

data mining concept associated with neural networks where predictions are made using a training data set

stop words

In database searching, "stop words" are small and frequently occurring words like and, or, in, of that are often ignored when keyed as search terms. Sometimes putting them in quotes " " will allow you to search them.

stemming

finding terms that share a common root and mean the same thing and combining them into one attribute

n-grams

a phrase or combination of words that may take one meaning that is different from or greater than the meaning of each owrd individually