Skip to content

🌱 Beginner Tier: Data Science Foundations

Levels 0–2 · Goal: fluency with data · ~10–14 weeks at about an hour a day

You start here if you know some Python but haven't done serious data work. By the end you'll be fast with the Python data stack, comfortable with the math that machine learning is built on, and able to run a complete data science project from a vague question to a clear, defensible answer.

What you'll be able to do

  • Set up reproducible Python environments and work productively in notebooks and scripts.
  • Manipulate data fluently with NumPy and pandas, and make charts that communicate clearly.
  • Read and use linear algebra, calculus, probability, and statistics, and check any formula numerically in code.
  • Query data with SQL, clean messy datasets, and explore them systematically.
  • Engineer features without leaking information from the future.
  • Design and analyze an A/B test correctly, and present findings to non-technical people.

The levels

  • Level 0: Python for data

    The data landscape, environments, Python essentials, NumPy, pandas, and visualization.

    Start Level 0

  • Level 1: Math foundations

    Linear algebra, matrix decompositions, calculus and gradients, probability, statistics, and information theory.

    Start Level 1

  • Level 2: Data science workflow

    SQL, data cleaning, exploratory analysis, feature engineering, A/B testing, and communicating results.

    Start Level 2

Tier checkpoint

You're ready for the Intermediate tier when you can do all of these without notes:

  • Explain NumPy broadcasting, and predict the shape of a[:, None] * b for any two arrays.
  • Answer a multi-step question about a dataset with a single chained pandas expression or SQL query with a window function.
  • Compute a gradient by hand and confirm it with a numerical check in NumPy.
  • Explain what a p-value is and isn't, and compute a confidence interval with the bootstrap.
  • Explain the SVD geometrically, and why it matters for data.
  • Spot data leakage in a feature engineering pipeline.
  • Size an A/B test with a power analysis, and explain why peeking at results inflates false positives.

See the full roadmap for every concept in this tier.