SET1

A machine learning model that can identify handwritten digits from images is an example of

• Unsupervised learning

• supervised learning

• Reinforcement learning

• Deep learning

What is the difference between simple linear regression and multiple linear regression?

• Simple linear regression uses categorical variables, while multiple linear regression uses continuous variables.

• Simple linear regression models the relationship between one independent variable and a dependent variable, while multiple linear regression models the relationship between two or more independent variables and a dependent variable.

• Simple linear regression is always more accurate than multiple linear regression.

• Multiple linear regression is a type of classification model, while simple linear regression is a regression model.

What is a common technique used to address overfitting in linear regression?

• • A) Increasing the number of independent variables in the model.

• • B) regularization (e.g., L1 or L2 regularization) which penalizes the model for having too many complex features.

• • C) Collecting more data.

• D) Both B and C.

In a confusion matrix, what does the term "True Positive" (TP) represent?

• • A) An instance correctly classified as negative.

• • B) An instance incorrectly classified as positive.

• • C) An instance correctly classified as positive.

• D) The total number of positive examples in the dataset.

2. Stratified sampling is particularly beneficial when dealing with datasets that are:

• • A) Perfectly balanced between classes.

• • B) Highly imbalanced, where some classes have significantly fewer examples than others.

• • C) Very large and require efficient sampling techniques.

• D) Composed entirely of continuous numerical features.

Entropy is a measure of

• • A. The accuracy of a decision tree.

• • B. The complexity of a decision tree.

• • C. The randomness or uncertainty in a dataset.

• D. The distance between two data points.

When making a split in a decision tree, the goal is to:

• • A. Increase the entropy of the resulting child nodes.

• • B. Maintain the same level of entropy in the child nodes.

• • C. Decrease the overall entropy of the data by separating the data points based on a feature.

• D. Choose the split that results in the most balanced child nodes.

Instance-based learning (IBL) algorithms are also known as

• • A. Generative learning models

• • B. Lazy learning models

• • C. Eager learning models

• D. Probabilistic learning models

KNN is a relatively simple algorithm. What is a potential drawback of this simplicity?

• • A. Requires a large amount of training data for good performance.

• • B. Can be computationally expensive for large datasets due to distance calculations for new data points.

• • C. Prone to overfitting if the training data is noisy

• • .

  • D. Difficult to interpret the reasons behind predictions.

Principal Component Analysis (PCA) is a popular dimensionality reduction technique. What does PCA do?

• • A. Identifies clusters of data points.

• • B. Projects data points onto a new set of axes (principal components) that capture most of the variance.

• • C. Selects a subset of the original features for further analysis.

• D. Transforms the data to a normal distribution.