AI & Data Science Mock Interview Questions & Practice

What is logistic regression, and why is it used for classification despite its name?

Key points to cover:

• Models the probability of a binary outcome
• Using a logistic (sigmoid) function
• Called "regression" because it models a continuous probability before thresholding

What is the difference between hierarchical clustering and K-means clustering?

Key points to cover:

• Hierarchical clustering builds a nested tree of clusters (dendrogram)
• Without needing to specify the number of clusters upfront
• K-means requires specifying the number of clusters (K) in advance

What are precision and recall, and how do they differ as classification metrics?

Key points to cover:

• Precision: proportion of predicted positives that are actually positive
• Recall: proportion of actual positives correctly identified
• They often trade off depending on the classification threshold

What is backpropagation, and what role does it play in training a neural network?

Key points to cover:

• Computes the gradient of the loss function with respect to each weight
• By propagating error backward from the output layer
• Used with an optimization algorithm to iteratively update weights

What is pooling in a CNN, and what is its purpose?

Key points to cover:

• Reduces the spatial dimensions of feature maps
• E.g. taking the max or average value in local regions
• Reduces computational cost, controls overfitting, provides translation invariance

What is the vanishing gradient problem in the context of standard RNNs, and how does it limit their practical use?

Key points to cover:

• Gradients shrink exponentially as they propagate backward
• Through many time steps
• Limits ability to learn long-range dependencies in sequences

What is Bayes' theorem, and what does it allow you to calculate?

Key points to cover:

• A formula relating conditional probabilities
• Allows calculating the probability of a hypothesis given evidence
• Based on the prior probability and the likelihood of the evidence

What is a normal (Gaussian) distribution, and what are its key characteristics?

Key points to cover:

• A continuous, symmetric, bell-shaped probability distribution
• Characterized by mean (center) and standard deviation (spread)
• The 68-95-99.7 rule for data within 1/2/3 standard deviations

What is a Type I error and a Type II error in hypothesis testing?

Key points to cover:

• Type I: incorrectly rejecting a true null hypothesis (false positive)
• Type II: incorrectly failing to reject a false null hypothesis (false negative)
• Notes both are trade-offs against each other

Explain how you would detect if two strings are anagrams of each other.

Key points to cover:

• Character frequency count comparison
• Sorting approach as an alternative, O(n log n)
• Discusses the time/space trade-off between the two approaches

How would you find the lowest common ancestor of two nodes in a binary search tree?

Key points to cover:

• Uses the BST property while traversing from the root
• States O(h) time complexity
• Contrasts with general binary-tree LCA needing a full traversal

How does quicksort's pivot selection affect its performance?

Key points to cover:

• Poor pivot choice (e.g. first element on sorted input) degrades to O(n^2)
• Randomized or median-of-three pivot selection helps avoid the worst case
• Connects pivot quality to partition balance

Explain the difference between multithreading and multiprocessing.

Key points to cover:

• Multithreading shares memory within one process, lower overhead
• Needs synchronization to avoid shared-state issues
• Multiprocessing uses separate memory spaces, higher isolation and overhead

Compare paging and segmentation as memory management schemes.

Key points to cover:

• Paging uses fixed-size units, transparent to the programmer
• Segmentation uses variable-size logical units (code/data/stack)
• Notes segmentation can suffer external fragmentation

How does a condition variable work, and what problem does it solve?

Key points to cover:

• Lets a thread wait/block until a specific condition becomes true
• Signaled by another thread
• Solves busy-waiting inefficiency compared to polling a flag

What is a subquery, and give an example of when you'd use one.

Key points to cover:

• A query nested inside another query
• Useful for filtering based on an aggregate or related table result
• Gives a concrete example (e.g. rows above a subquery-computed average)

Explain Third Normal Form (3NF) and what problem it solves.

Key points to cover:

• 2NF plus no transitive dependency
• Non-key columns must depend only on the primary key
• Removes redundancy caused by indirect (transitive) dependencies

Explain the difference between the READ COMMITTED and REPEATABLE READ isolation levels.

Key points to cover:

• READ COMMITTED prevents dirty reads
• READ COMMITTED still allows non-repeatable reads within a transaction
• REPEATABLE READ additionally prevents non-repeatable reads