In The Past
ML
Dreaded Words
A and B
In The Attic
100

This regression coefficient tells the expected change in the outcome for a one-unit increase in a predictor, all else equal in the model.

What is a slope or BETA coefficient?

100

This learning setup trains a model using examples that already include the correct outcome labels.

What is supervised learning?

100

How close you are to the real answer

Accuracy

100

Does correlation implies causality?

False

100

The method for resampling of an existing sample dataset

Bootstrapping

200

This estimation method chooses coefficients by making the squared prediction errors as small as possible.

What is ordinary least squares, or OLS?

200

This problem happens when a model learns noise in the training data and performs worse on new cases.

What is overfitting?

200

You are actually measuring or testing something that you said you were going to test

Validity

200

Does causality imply correlation

True

200

Time series data can be decomposed into three elements, alpha, beta, gamma

Base (history), trend, seasonality

300

This type of variable lets a regression include categories such as region, gender, or treatment group numerically.

What is a dummy variable?

300

This ensemble method builds many trees on varied samples and features, then combines them to reduce variance.

What is a random forest? NOT on test!

300

You measure, then remeasure, and they return the same results

Reliability

300

Significance in the A/B test can be determined using this simple test

T-test

300

ARIMA

Autoregressive Integrated Moving Average

400

This problem occurs when a predictor is related to the unobserved error term, threatening unbiased causal interpretation.

What is endogeneity?

400

Type of multiple regression that is also a machine learning model.

Multiple logistic regression, multiple BOOTSTRAP regression, etc...

400

Someone else testing your results and obtain similar answers

Replicability

400

Causal inference approach using a modified regression, where we see if the changes are "caused" by some policy change

Difference-in-Differences

400

Used to model and forecast volatility

GARCH

500

This approach uses an external variable related to the endogenous predictor but not directly to the outcome except through that predictor.

What are instrumental variables or two-stage least squares?

500

Tree splitting approach used in prediction...

Regression trees or binomial trees (next class)

500

One group of students in one class returns the same results as another class within the same university 

Consistency

500

The causality model named after a famous Nobel laureate

Granger causality

500

Seven days a week can be modeled using this many dummies

6

M
e
n
u