stud
bud
Sign up
Log in
Home
Library
Explore
SAT
More
Day 3 Flashcards | StudBud
Explore
/
Day 3
Day 3
24 terms
·
StudBud Administration
·
34 plays
Imported from external platform
Flashcards
24 cards
F
Practice quiz
P
Host live
L
Questions
24
Answers
1
Given the loss function of MSE what is our goal?
Find the slope and intercept which minimize MSE
idk
Find the slope and intercept which maximize MSE
Plug in slope and intercept to find loss
2
During model development, which dataset is used to tune hyperparameters and select the best model?
Deployment
Validation
Training
Testing
3
In supervised learning, which task involves forecasting a discrete category label?
Clustering
Regression
Density estimation
Classification
4
In the simple linear model f\left(X\right)+ε what is the form of f\left(X\right)?
β_1X+β_0
β_1+β_0X+ε
β_0X^2+ε
β_0X+ε
5
During which stage do you handle missing values, correct errors, and transform variable?
Exploration of intrinsic data structures
Data cleaning and pre-processing
Data Collection
Predictive and inferential analysis
6
Is this a linear model?
No
Yes
7
What is the correct order of datasets in the typical model-building flow?
Training → Test → Validation
Training → Validation → Test
Test → Validation → Training
Validation → Training → Test
8
Which framework makes optimizing a multiple linear regression problem simpler?
Writing the inputs in vector notation
Multiple partial derivatives
Plotting the function on a multiple dimension graph
9
What loss function do we use in Ordinary Least Squares?
Hinge loss
Mean Absolute Error
Cross-entropy
Mean Squared Error
10
What name is given to the variable whose value we want to predict?
Feature
Hyperparameter
Predictor variable
Response variable
11
What does "overfitting" refer to?
Your model has really high accuracy on training data but not test data
Your model has really high accuracy on training data
Your model has really high accuracy on test data but not training data
Your model has really high accuracy on test data
12
Which dataset is held out until the very end to provide an unbiased evaluation of model performance?
Deployment
Testing
Validation
Training
13
What benefits does polynomial regression provide?
Allows for more complex relationships to be modeled
It is easier to minimize the MSE loss function
Allows you to fit a model to more data
Allows for multiple input variables to be modeled
14
What is multiple linear regression?
Optimizing a MSE with many different Y output values
Optimizing a MSE with many different X inputs
Trying to optimize many different MSE functions
When the optimization problem has multiple steps
15
What is the primary objective of predictive analysis in the Data Science Life Cycle?
Group data into clusters
Communicate findings to stakeholders
Use past or current data to predict future unseen outcomes
Estimate probability distributions of data
16
The purpose of including terms in your regression model like the following is to:
To model the relationships input variables may have
To model quadratic output behaviors
To add non-linearity
17
What name is given to the variables whose values we use to make our prediction?
Residuals
Features
Responses
Targets
18
What method of differential calculus is used to find the minimum of MSE?
Partial derivatives
Integral
Taylor series
Derivatives
19
What is the very first stage of the Data Science Life Cycle?
Data pre-processing/cleaning
Exploratory Data Analysis
Domain Problem Definition
Predictive and inferential analysis
20
Which core ML problem estimates a function to describe the underlying data distribution?
Regression
Classification
Clustering
Density estimation
21
In simple linear regression the parameter β_0 represents:
The slope
The intercept
Predictor
Error term
22
What method is most commonly used to minimizing the MSE loss function?
Testing out many values of slope and intercept to find minimum output
Guess
Use differential calculus to find the minimum
Graphing the function and seeing which one looks close
23
Which risk increases as you fit higher‐degree polynomials to data?
Harder optimization problem
Underfitting
Larger MSE
Overfitting
24
If your goal is
not
to predict or forecast a value, which branch of learning do you follow?
Supervised learning
Reinforcement learning
Semi-supervised learning
Unsupervised learning
