ML
ML Interview Roadmap
← Study plan·Math & Stats

Day 9 of 133

Eigenvalues & condition number + DSA Math & Geometry

Eigen-decompositions, spectral theorem, why condition number matters in training.

DSA · NeetCode Math & Geometry

  • Happy NumberDSA · Math & Geometry

    Interview questions to prep

    1. Where does integer overflow / negative input / zero hide here, and how do you guard against it?
    2. Can you derive a closed-form solution, and how does it compare to the iterative one?
    3. Walk through edge cases: 0, 1, max int, min int, negative input.
  • Plus OneDSA · Math & Geometry

    Interview questions to prep

    1. Where does integer overflow / negative input / zero hide here, and how do you guard against it?
    2. Can you derive a closed-form solution, and how does it compare to the iterative one?
    3. Walk through edge cases: 0, 1, max int, min int, negative input.
  • Powx NDSA · Math & Geometry

    Interview questions to prep

    1. Where does integer overflow / negative input / zero hide here, and how do you guard against it?
    2. Can you derive a closed-form solution, and how does it compare to the iterative one?
    3. Walk through edge cases: 0, 1, max int, min int, negative input.

Math · Eigenvalues, eigenvectors, decompositions

  • Eigenvalues & eigenvectors — the geometric pictureStatistics3Blue1Brown

    Interview questions to prep

    1. What does it mean intuitively for a vector to be an eigenvector of a matrix?
    2. Where do eigenvalues show up in ML?
  • Spectral theorem & symmetric matricesStatisticsRead

    Interview questions to prep

    1. Why are eigenvectors of symmetric matrices orthogonal, and why does that matter for PCA?
    2. Compare diagonalization vs eigendecomposition.
  • Condition number & numerical stabilityStatisticsRead

    Interview questions to prep

    1. What is the condition number of a matrix and why does a high condition number hurt training?
    2. Give two practical fixes when your design matrix is ill-conditioned (e.g., regularization, feature scaling, removing collinear features).

References & further reading