FastPrepImplement K-Means Clustering
Problems/Implement K-Means Clustering

Implement K-Means Clustering

Apple logoApple● HardNEW GRADPHONE SCREEN

Updated Nov 2026

QuestionFunctionExamplesConstraints

Problem statement

Source note: Reported in an Apple new grad phone screen. The report asks for K-Means. It does not say how to pick the starting points, when to stop, or how to break ties. We chose those. Our version is about 60% the same as the reported question.

Implement K-Means clustering without a clustering library. There are no random choices. Initialize the k centroids from the first k points, in input order. Duplicate initial centroids are allowed.

Perform at most rounds assignment passes. In each pass:

  1. For each point, compare squared Euclidean distances to the current centroids. Scan cluster indices from 0 upward. Replace the current best only if a later distance is smaller by more than 10^-9; otherwise keep the earlier index.
  2. If all labels equal those from the previous pass, stop.
  3. Replace each nonempty cluster's centroid by the coordinate-wise mean of its assigned points. Keep an empty cluster's previous centroid.

Return each point's zero-based cluster index from the last assignment pass. Do not run an extra assignment after the pass limit. Keep the input points unchanged.

Function

clusterPoints(points: int[][], k: int, rounds: int) → int[]

Examples

Example 1

points = [[0,0],[10,10],[1,0],[9,10]]k = 2rounds = 10return = [0,1,0,1]

The first two points initialize centroids 0 and 1. The two points near the origin join 0, and the two near (10,10) join 1. Updated centroids (0.5,0) and (9.5,10) preserve these assignments.

Example 2

points = [[0],[4],[2]]k = 2rounds = 1return = [0,1,0]

On the single assignment pass, point 2 is equally distant from centroids 0 and 4. The smaller cluster index wins, so it joins cluster 0.

Constraints

  • 1 <= points.length <= 100
  • Each point has 1 to 5 coordinates.
  • -100 <= points[i][j] <= 100
  • 1 <= k <= 10
  • k <= points.length
  • 1 <= rounds <= 50
  • All points have the same dimension.

More Apple Phone Screen questions

  • Subarray Product Less Than KPHONE SCREEN · Seen Nov 2026
  • Evaluate DivisionPHONE SCREEN · Seen Nov 2026
  • Reverse a Linked ListPHONE SCREEN · Seen Nov 2026
  • Validate a BST with a Custom ComparisonPHONE SCREEN · Seen Nov 2026
  • Design a Food Rating SystemPHONE SCREEN · Seen Oct 2026
  • Days Until Poisoned Plants StabilizePHONE SCREEN · Seen Oct 2026
  • Capture Surrounded Regions with Union-FindPHONE SCREEN · Seen Oct 2026
  • Maintain a Consistent-Hashing RingPHONE SCREEN · Seen Oct 2026
  • Count Nodes in a ChainPHONE SCREEN · Seen Oct 2026
  • Run-Length DecodingPHONE SCREEN · Seen Oct 2026

Similar questions at other companies

  • Deterministic K-Means AssignmentsLyft · PHONE SCREENLyft interview questions
  • Online Chess Move ValidationOpenAI · PHONE SCREENOpenAI interview questions
  • Deterministic K-Means AssignmentsMeta · PHONE SCREEN · ONSITE INTERVIEWMeta interview questions
  • Knockout Tournament Win ProbabilityGoogle · PHONE SCREENGoogle interview questions
  • Scaled Dot-Product AttentionAmazon · PHONE SCREEN
  • Deterministic K-Means AssignmentsMicrosoft · PHONE SCREEN · ONSITE INTERVIEWMicrosoft interview questions
  • Automation Pipeline, Part 1: Single WorkerPlaid · PHONE SCREENPlaid interview questions
  • Constant-Time Weighted Pick with an Alias TableEtched · PHONE SCREEN
  • Apply Affine Array TransformationsByteDance · PHONE SCREENByteDance interview questions
  • Exponential Backoff DelaysWonderful · PHONE SCREEN

Apple interview experiences

  • Apple SWE Phone Screen: Subarray Products: 1 Experience
  • Apple New Grad SWE Java, Coding and Manager Rounds: 1 Experience
  • Apple AI Coding and Related Interviews: 17 Experiences
  • Apple India SWE New Grad Interviews: 5 Experiences
  • Apple New Grad SWE Phone Screens: 12 Experiences (Part 3 of 3)
  • Apple New Grad SWE and Embedded: 16 Experiences (Part 2 of 3)
All Apple interview questionsSee Apple hiring insights
Saved
class Solution {
    public int[] clusterPoints(int[][] points, int k, int rounds) {
        // Write your code here.
    }
}
Checking account…