WorksheetsMAchine-learning-2
Total questions: 20
Worksheet time: 13mins
Increasing λ (lambda) in Ridge Regression will
A) Increase coefficient magnitudes
B) Shrink coefficients toward zero
C) Make coefficients exactly zero
D) Have no effect
Learning rate too high in GD may cause:
A) Slow convergence
B) Oscillation around minimum
C) Perfect convergence
D) No change
One step of gradient descent with MSE loss involves updating:
A) Only bias
B) Only weights
C) Both weights and bias
D) Only λ
High bias leads to: underfitting or overfitting?
(a)
R² can never be:
A) 1
B) 0
C) -0.5
D) 0.7
Spam email detection is an example of:
A) Regression
B) Classification
C) Clustering
D) Dimensionality reduction
Grouping customers based on buying behavior without predefined categories is:
A) Regression
B) Classification
C) Clustering
D) Association Rule Mining
Answer (one step gradient Descent step with MAE):
Given y={(0,1),(1,3),(2,5)} m=0.5 b=0.5, α=0.05
m=0.55 b=0.55
m=0.55 b=0.45
m=0.45 b=0.55
m=0.45 b=0.45
apply normalization on x={10,11,12,13,200} answer should include 4 decimal numbers
(a)
Polynomial regression is useful when:
A) Data shows a linear trend
B) Data shows a non-linear trend
C) There is only one feature
D) No features are correlated
Overfitting in polynomial regression usually occurs when:
A) The degree is too low
B) The degree is too high
C) Learning rate is high
D) Number of data points is very large
Compute the slope m of the best-fit line using linear regression formula for x=[1,2,3],y=[2,4,6]
(a)
With one features (x1) and total polynomial degree M=6 (with all cross-terms, plus bias), how many coefficients are there?
(a)
A regression model has features x1,x2,x3,x4 and output y. Write the full linear model including bias, and indicate which symbols represent:
(a)
x=[0,1,2,3,4],y=[1,3,7,13,21]
Which regression model is more suitable?
Linear
Polynomial
Training a self-driving car to navigate a track by maximizing score is an example of:
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Semi-supervised Learning
Linear regression can model non-linear data accurately if you apply a polynomial feature transformation. True/False
(a)
If the training error is low but test error is high in polynomial regression, the model is:
A) Underfitting
B) Overfitting
C) Perfect fit
D) Linear
You have a dataset with many features, and you want feature selection while reducing overfitting. Which method is preferred?
A) Lasso Regression
B) Ridge Regression
C) Linear Regression
D) None
You have a dataset where all features are somewhat important, and you want to reduce model complexity but not eliminate features. Which method is preferred?
A) Lasso Regression
B) Ridge Regression
C) Linear Regression
D) Elastic Net
