How do I practice coding and algorithm questions?

Use PracHub's coding console to write, test, and debug your solutions in Python or JavaScript. View hints, test against sample inputs, and compare with official solutions.

What difficulty level is this coding question?

This is a medium difficulty Coding & Algorithms question, commonly asked during Onsite rounds at Amazon.

What role is this question designed for?

This question is commonly asked for Machine Learning Engineer candidates at Amazon during technical interviews.

Implement K-means and solve interval/frequency tasks

Quick Overview

This multi-part question evaluates understanding of K-means clustering, interval-merging algorithms, and frequency-based top-k retrieval, assessing competencies in unsupervised learning concepts, algorithmic problem solving, data structures, and complexity analysis.

Company: Amazon

Role: Machine Learning Engineer

Category: Coding & Algorithms

Difficulty: medium

Interview Round: Onsite

## Task 1 — Describe/implement K-means clustering Given: - A data matrix `X` with shape `(n_samples, d)`. - An integer `k` (number of clusters). Explain (or write high-level pseudocode for) the K-means algorithm such that: - The steps are unambiguous. - Matrix/vector dimensions are consistent. - You cover initialization, assignment, centroid update, and stopping criteria. ## Task 2 — Merge overlapping intervals Given a list of closed intervals `intervals`, where each interval is `[start, end]` and `start <= end`, merge all overlapping intervals and return the merged list. **Input:** `intervals: List[List[int]]` **Output:** merged intervals in any order (or sorted by start), with no overlaps. ## Task 3 — Return the top-k most frequent elements Given an integer array `nums` and an integer `k`, return the `k` values that occur most frequently. **Input:** `nums: List[int]`, `k: int` **Output:** `List[int]` containing `k` elements. ### Constraints (assume typical interview constraints) - `1 <= n <= 2e5` - Values fit in 32-bit signed integers - `1 <= k <= number of distinct values in nums`

Quick Answer: This multi-part question evaluates understanding of K-means clustering, interval-merging algorithms, and frequency-based top-k retrieval, assessing competencies in unsupervised learning concepts, algorithmic problem solving, data structures, and complexity analysis.

Part 1: Deterministic K-means Clustering

Given a data matrix X with n points and d features per point, implement a deterministic version of K-means clustering. Use the first k points of X as the initial centroids. In each iteration, assign every point to the nearest centroid using squared Euclidean distance; if two centroids are tied, choose the smaller centroid index. Then update each centroid to the coordinate-wise mean of the points assigned to it. If a cluster receives no points, keep its centroid unchanged. Stop when the labels no longer change or when max_iters iterations have been executed. Return both the final centroids and the label for each point. Do not round the centroid values.

Constraints

1 <= n <= 10^3
1 <= d <= 20
1 <= k <= n
1 <= max_iters <= 100
All rows of X have the same length d
Each coordinate fits in standard Python int/float ranges

Examples

Input: ([[1, 1], [1, 2], [4, 4], [4, 5]], 2)

Expected Output: {'centroids': [[1.0, 1.5], [4.0, 4.5]], 'labels': [0, 0, 1, 1]}

Explanation: Two obvious groups form, and the centroids converge to the means of those groups.

Input: ([[0, 0], [2, 2], [4, 4]], 1)

Expected Output: {'centroids': [[2.0, 2.0]], 'labels': [0, 0, 0]}

Explanation: With one cluster, the centroid becomes the mean of all points.

Input: ([[0], [0], [10]], 2)

Expected Output: {'centroids': [[10.0], [0.0]], 'labels': [1, 1, 0]}

Explanation: The initial centroids are identical, so one cluster is empty in the first update and must remain unchanged.

Input: ([[5], [9], [13]], 3)

Expected Output: {'centroids': [[5.0], [9.0], [13.0]], 'labels': [0, 1, 2]}

Explanation: When k equals the number of points, each point remains its own cluster.

Hints

Use squared Euclidean distance so you do not need to call sqrt.
During the update step, keep a sum vector and a count for each cluster; if a count is 0, leave that centroid unchanged.

Part 2: Merge Overlapping Closed Intervals

You are given a list of closed intervals, where each interval is [start, end] and start <= end. Merge all overlapping intervals and return a list of disjoint intervals sorted by starting value. Because the intervals are closed, [1, 4] and [4, 5] should be merged.

Constraints

0 <= len(intervals) <= 2 * 10^5
-2^31 <= start <= end <= 2^31 - 1

Examples

Input: ([[1, 3], [2, 6], [8, 10], [15, 18]],)

Expected Output: [[1, 6], [8, 10], [15, 18]]

Explanation: The first two intervals overlap and become [1, 6].

Input: ([[1, 4], [4, 5]],)

Expected Output: [[1, 5]]

Explanation: Closed intervals that touch at an endpoint still overlap.

Input: ([],)

Expected Output: []

Explanation: An empty input has no intervals to merge.

Input: ([[6, 8], [1, 9], [2, 4], [4, 7]],)

Expected Output: [[1, 9]]

Explanation: After sorting, every interval overlaps with [1, 9].

Hints

Sort the intervals by their start value first.
As you scan left to right, only compare the current interval with the last merged interval.

Part 3: Top-k Most Frequent Elements with Deterministic Tie-Breaking

Given an integer array nums and an integer k, return the k values that occur most frequently. To make the result deterministic, order the answer by decreasing frequency. If two values have the same frequency, the smaller numeric value must come first.

Constraints

1 <= len(nums) <= 2 * 10^5
Values fit in 32-bit signed integers
1 <= k <= number of distinct values in nums

Examples

Input: ([1, 1, 1, 2, 2, 3], 2)

Expected Output: [1, 2]

Explanation: 1 appears 3 times, 2 appears 2 times, and 3 appears once.

Input: ([4, 4, 1, 1, 2, 2], 2)

Expected Output: [1, 2]

Explanation: All three values appear twice, so the smaller values come first.

Input: ([7], 1)

Expected Output: [7]

Explanation: A single-element array returns that element.

Input: ([-1, -1, -2, -2, -3], 2)

Expected Output: [-2, -1]

Explanation: -1 and -2 are tied in frequency, so the smaller number -2 comes first.

Hints

Count how many times each value appears before trying to choose the top k.
Once you know the frequencies, sort distinct values by frequency descending and value ascending.

Quick Overview