Crowdly

Add to Chrome

Machine Learning: Theory and Practice-Lecture,Section-1-Fall 2025

Looking for Machine Learning: Theory and Practice-Lecture,Section-1-Fall 2025 test answers and solutions? Browse our comprehensive collection of verified answers for Machine Learning: Theory and Practice-Lecture,Section-1-Fall 2025 at moodle.nu.edu.kz.

Get instant access to accurate answers and detailed explanations for your course questions. Our community-driven platform helps students succeed!

The Iris dataset has 3 species and 4 features: sepal length, sepal width, petal length, and petal width. Analysis shows:

Petal length and petal width are strongly positively correlated.

Sepal length and petal length are moderately correlated.

Based on this, which statements are correct?

A) Including both petal length and petal width may introduce redundancy and does not necessarily improve model performance.

B) For Linear Discriminant Analysis (LDA), highly correlated features can cause the within-class covariance matrix to become ill-conditioned, reducing model stability.

C) Moderate correlation between sepal length and petal length means these features are completely independent.

D) Strongly correlated features are always preferred for any machine learning model because they increase predictive power.

All of the above

B and C

A and B

A, C, and D

View this question

You are designing a K-NN classifier for a 2-class problem. Which of the following is the best practice for choosing k and initializing the classifier in Python?

Choose k randomly;

from sklearn.linear_model import LogisticRegression

model = LogisticRegression()

Choose an odd k to reduce tie votes;

from sklearn.neighbors import KNeighborsClassifier

model = KNeighborsClassifier(n_neighbors=5)

100%

Choose k as a multiple of number of classes;

from sklearn.svm import SVC

model = SVC()

Choose an even k to balance votes;

from sklearn.neighbors import KNeighborsClassifier

model = KNeighborsClassifier(n_neighbors=4)

View this question

Consider the standard perceptron training algorithm with training set learning rate α, and maximum epochs (T). The weight update rule is:

The perceptron updates weights only when the prediction is correct.

The perceptron converges in a finite number of steps if the data is linearly separable.

The learning rate (α) must decrease over time to guarantee convergence.

The bias term (b) is updated independently of the weight vector ().

View this question

Why do we need Gradient Descent in machine learning?

Because it provides closed-form solutions for all models.

Because it avoids the use of GPUs and vectorized operations.

Because it is only useful for linear regression problems.

Because it is iterative, scales to large datasets, and works with many differentiable loss functions.

View this question

You are training a house price prediction model with 5,000 training examples using Mini-Batch Gradient Descent (MBGD) of batch size = 32. Which of the following statements correctly describes how MBGD works and why it is widely used?

MBGD performs about 156 updates per epoch, each using 32 houses, balancing efficiency (via vectorization) and stability (less noisy than SGD).

100%

MBGD does not require a learning rate hyperparameter, since averaging over mini-batches eliminates the need for step size control.

MBGD performs only 1 update per epoch, using all 5,000 houses at once, giving deterministic but very slow convergence.

MBGD performs 5,000 updates per epoch, one for each training house, giving fast progress but highly noisy convergence.

View this question

Given a feature x with values in the range [300,2000], different scaling methods are applied. Which of the following correctly represents Z-score Normalization?

100%

View this question

Day	Outlook	Temperature	Play Tennis
1	Sunny	Hot	No
2	Sunny	Mild	Yes
3	Overcast	Hot	Yes
4	Rainy	Mild	Yes
5	Sunny	Cool	No
6	Rainy	Cool	Yes

If the Outlook is Sunny, what is the probability that the player will Play Tennis (Yes)?

100%

View this question

Which of the following correctly lists the sequence of K-NN steps and the Python command to import and initialize the K-NN classifier?

Sequence: Aggregate labels → Select neighbors → Compute distances → Choose k

Command: from sklearn.models import KNeighborsClassifier

Sequence: Compute distances → Choose k → Aggregate labels → Select neighbors

Command: from sklearn.linear_model import KNN

Sequence: Choose k → Aggregate labels → Compute distances → Select neighbors

Command: from sklearn.neighbors import KNearestNeighbors

Sequence: Choose k → Compute distances → Select neighbors → Aggregate labels

Command: from sklearn.neighbors import KNeighborsClassifier

100%

View this question

Match each mathematical form with the correct Python implementation.

View this question

You have a classifier and predictions y_pred for your test set y_test. You want to compute and visualize the confusion matrix using Python. Which of the following options is correct?

from sklearn.metrics import confusion_matrix

import seaborn as sns

cm = confusion_matrix(y_test, y_pred)

sns.heatmap(cm, annot=True, fmt='d', cmap='Blues')

from sklearn.metrics import accuracy_score

import seaborn as sns

cm = accuracy_score(y_test, y_pred)

sns.heatmap(cm, annot=True, fmt='d')

from sklearn.metrics import confusion_matrix

import seaborn as sns

cm = confusion_matrix(y_pred, y_test)

sns.heatmap(cm, annot=True, fmt='d', cmap='Blues')