IBM AI Engineering Machine Learning with Python

1. In a decision tree, what is a 'split' based on?

Feature threshold that best separates classes

Random feature selection

Number of samples

Target variable mean

2. In unsupervised learning, what is the primary goal?

Discover patterns and structure in unlabeled data

Predict target variable values

Maximize classification accuracy

Minimize prediction error

3. What is the main advantage of using Random Forest over a single decision tree?

Reduces overfitting through ensemble averaging

Requires less computational power

Eliminates need for feature scaling

Automatically handles missing values

4. In gradient descent, what does the learning rate control?

Step size for parameter updates

Number of iterations needed

Feature scaling magnitude

Training data split ratio

5. Which algorithm is most suitable for non-linear classification problems?

Support Vector Machine with RBF kernel

Linear regression

Logistic regression

Simple linear classifier

6. What does 'precision' measure in classification?

Proportion of true positives among predicted positives

Proportion of true positives among actual positives

Total correct predictions

Balance between true and false positives

7. In logistic regression, what activation function is used for binary classification?

Sigmoid function

ReLU function

Tanh function

Linear function

8. Which pandas method is used to handle missing values by replacing them with a specified value?

Fillna()

Dropna()

Replace()

Interpolate()

9. What is the purpose of feature engineering in machine learning?

Create meaningful features to improve model performance

Reduce dataset size

Eliminate all categorical variables

Guarantee model accuracy

10. Which ensemble method combines weak learners sequentially, with each correcting previous errors?

Boosting

Bagging

Stacking

Voting

11. Which Python library is most commonly used for numerical computations and array operations in machine learning?

NumPy

Matplotlib

Flask

Selenium

12. What does cross-validation help prevent in model development?

Overfitting and biased performance estimates

Feature scaling issues

Missing values

Duplicate records

13. Which regularization technique adds a penalty term proportional to the square of coefficients?

Ridge regression (L2)

Lasso regression (L1)

Elastic Net

Dropout

14. What is the elbow method used for in clustering?

Determining optimal number of clusters

Scaling feature values

Removing outliers

Selecting classification algorithm

15. In k-means clustering, what determines the initial cluster centers?

Random selection or k-means++ initialization

The mean of all data points

The largest data points

User-defined coordinates only

16. Which scikit-learn class is used for standardizing features to have mean 0 and standard deviation 1?

StandardScaler

MinMaxScaler

RobustScaler

PolynomialFeatures

17. What does 'overfitting' mean in machine learning?

Model learns training data too well, performs poorly on new data

Model fails to capture patterns in training data

Model requires too many features

Model has insufficient parameters

18. Which metric is best for evaluating classification models with imbalanced datasets?

Accuracy

F1-score

Mean squared error

R-squared

19. What is the primary purpose of train-test splitting in machine learning?

To evaluate model performance on unseen data

To increase dataset size

To remove outliers

To normalize feature values

20. In supervised learning, what role does the target variable play?

It provides labels for training the model

It reduces computational complexity

It eliminates the need for feature scaling

It automatically selects features

IBM AI Engineering Machine Learning with Python Quiz