Font size
WorksheetsLogistic Regression and k-NN
Total questions: 20
Worksheet time: 7mins
Logistic Regression is mainly used for:
Feature selection
Classification problems
Clustering problems
Regression problems
Logistic Regression uses (a) function to map values into probabilities.
(ans : all small case letters)
In k-NN, “k” refers to:
Number of clusters
Number of training samples
Number of nearest neighbors considered
Number of features
(a) distance metric is most commonly used in k-NN.
(ans : in small case)
If k in k-NN is too small, the model may:
(ans : in small case)
(a)
k-NN is considered a:
Parametric model
Generative model
Probabilistic model
Non-parametric model
In k-NN, when classes are imbalanced, the best solution is:
Use weighted voting
Remove features
Use smaller k
Use sigmoid activation
The cost function used in Logistic Regression is:
Hinge loss
Euclidean distance
Cross-Entropy
Mean Squared Error (MSE)
k-NN works best when:
Data is very high-dimensional
Data has many missing values
Features are categorical only
Data is low-dimensional and well-separated
Which of the following is TRUE about k-NN?
Training is slow, prediction is fast
Training is fast, prediction is slow
Both training and prediction are slow
Both training and prediction are fast
Logistic Regression uses (a) to estimate parameters.
(answer in short form : give all in capital letters)
The output of Logistic Regression is interpreted as (a) of belonging to a class.
(ans : all small case)
The range of the sigmoid function in Logistic Regression is:
ans : provide in format "start_range , end_range"
(a)
k-NN is considered a:
Clustering algorithm
Parametric model
Non-parametric model
Semi-supervised model
The logit function in Logistic Regression is defined as:
(function : give in all small case letters without space)
(a)
Which of the following is NOT true about k-NN?
It needs feature scaling
It requires training time
It predicts based on majority vote
It is a lazy learner
Which is a disadvantage of Logistic Regression?
Cannot handle binary classification
Cannot be extended to multi-class problems
Cannot interpret coefficients
Requires features to be linearly separable
k-NN suffers when the dataset has:
Small number of features
Balanced data
Large number of features
Normalized data
Which of the following is TRUE about Logistic Regression?
It can predict probabilities
It only works with categorical input features
It is non-parametric
It doesn’t require optimization
Logistic Regression can be used for:
Predicting stock prices
Predicting exam scores
Predicting whether a student passes/fails
All of the above
