Learning archive · Educational notebook

Computer Performance Prediction

A regression notebook with an explicit holdout.

An educational random-forest regression study of hardware performance using six predictors.

The question

Prediction error needs both a clear unit and an explicit held-out population to be interpretable.

How it works

  • Model the recorded hardware-performance target from six predictors.
  • Separate a 47-system holdout from the 208-record dataset.
  • Inspect regression error alongside explained variation.

Evidence & results

  • Stored model evaluation reports test R² of 0.948 on the 47-system holdout.

Scope & next steps

  • Historical educational result from one notebook, not a production capacity model.
  • The reported estimator is separate from the randomized-search estimator. This shares a repository with Mushroom Classification.

Go one level deeper.

Ask about the implementation, tradeoffs, or my contribution.

Explore with the portfolio guide