Quiz 6 Jeopardy (part B)

Acronyms

200

MCMC

Markov Chain Monte Carlo

400

NUTS

No-U-Turn Sampler

600

ARIMA

Auto Regressive Integrated Moving Average

800

TFIDF

Term Frequency–Inverse Document Frequency

1000

LASSO

Least Absolute Shrinkage and Selection Operator

Statistics

200

This is a type of estimate computed from the statistics of the observed data and proposes a range of plausible values for an unknown parameter.

Confidence Interval

400

For a continuous random variable, this provides the likelihood the random variable is equal to the value given

Probability Density Function (PDF)

600

In Bayesian Inference, this component is the probability distribution of the observed data and serves as a normalizing agent

Marginal Distribution

800

Singular value decomposition (used in PCA) has the following equation:

A = USV^T

The first three values (A, U, & S) represent these matrices in order.

1) A = Covariance Matrix

2) U = Eigenvectors (or Principle Components)

3) S = Eigenvalues

1000

In order to plot the ROC Curve, these two metrics are used for the x and y axis, respectively.

False Positive Rate (or 1 - Specificity) and Sensitivity (or Recall or True Positive Rate)

Data Science Movies

200

This popular Japanese anime series features aliens, androids, and wish granting magical creatures spawned from balls. With so much going on, one might wish things not go too many standard deviations from the mean.

Dragonball Z-score

(Dragonball Z + Z-score)

400

This 1987 comedy features Danny DeVito and Billy Crystal who plan to "swap murders", culminating in a final attempt to toss a mother onto a railway. Perhaps if they divide the data another way, the plan would have succeeded.

Throw Mama from the Train/Test Split

(Throw Mama from the Train + Train/Test Split)

600

This epic Sci-Fi/Action film is part of a trilogy (originally) with people revealing their fatherhood and deal's getting worse all the time. With so many problems near the end, one might want to return to the beginning, update the weights, and repeat the process until the problems aren't so bad.

Star Wars Episode V: The Empire Strikes Back Propagation

(Star Wars Episode V: The Empire Strikes Back + Back Propagation)

800

This Shrek spin off features a stylish cat voiced by Antonio Banderas. It was also generated by resampling randomly with replacement.

Puss in Bootstrap(ping)

(Puss in boots + Bootstrap(ping))

1000

This Sci-Fi thriller stars Jake Gyllenhaal as a troubled teen experiencing doomsday-related visions, often alongside a creepy bunny mask and the song "Mad World". I'm sure finding a connection between two random variables can do that to a person.

Donnie Darko-rrelation

(Donnie Darko + Correlation)

Python Trivia

200

This is the name of a function that belongs to an object

Method

400

This is the result from the following code execution:

SyntaxError: positional argument follows keyword argument

600

This is the term for when things within a function can only be accessed from within the function

Local Variables (Scope)

800

This is the term for when a method returns it's associated object allowing for additional method calls.

For example:

df.dropna().describe().round(2)

(Method) Chaining

1000

This is the result of the following code execution:

'a' + 'x' if '123'.isdigit() else 'y' + 'b'

'ax'

The conditional expression has a lower precedence than other operators, so the + operator binds more tightly. It's equivalent to the following:

('a' + 'x') if '123'.isdigit() else ('y' + 'b')

Model Models

200

This model creates new features in order to create linear separability.

Support Vector Machines (SVMs)

400

These are the gini impurities for the following two nodes, respectively:

NODE 1: 400 class A, 0 class B, 400 class C

NODE 2: 10 class A, 10 class B, 10 class C

NODE 1: 50% impurity

NODE 2: 66.67% impurity 

600

This is the proper interpretation for the following logistic regression coefficent for income:

For every 1 unit increase in income, an individual is 1.01 times as likely to have a mansion

OR

For every 1 unit increase in income, an individual is 1% more likely to have a mansion

800

In a Convolutional Neural Network, this step must be done before feeding the image data into the first Dense layer.

Flatten

1000

In the AdaBoost model, this classification metric is used to determine the weight for each model in the ensemble, modifying how each model contributes to the final vote.

Missclassification Rate