Learn to inspect, visualize, and reason about data before training models.
Data and mathematics
Status: Data track in progress. Start with the linked lessons below.
What students will produce
- a documented dataset with a schema, units, and data-quality notes;
- charts that separate observations from explanations;
- a short analysis that reports uncertainty and limitations; and
- a reproducible notebook that another student can run.
Ordered subtopics
- Data types, tables, files, and schemas
- NumPy arrays and vector operations
- pandas tables and data cleaning
- Charts and exploratory analysis
- Mean, median, variance, and outliers
- Probability, sampling, and uncertainty
- Correlation, causation, and data bias
- Vectors, matrices, and dot products
- Optimization and gradients
- Reproducible experiments and dataset documentation
Planned project: analyze a public dataset and publish a short data story with a documented question, chart, limitation, and conclusion.
Best practices
- Keep raw data separate from cleaned data.
- Record units, missing-value decisions, and the source of every dataset.
- Never infer causation from a chart or correlation alone.
- Split data before fitting transformations used by a model.
- Prefer a simple, interpretable analysis before adding complexity.