mlbeginner
Precision and Recall Explained
Learn what precision and recall mean and why the right metric depends on the real-world cost of errors.
Mar 25, 2026•1 min read
mlbeginner
Train Test Split for Beginners
Understand why machine learning needs a train/test split and what goes wrong when you skip it.
Mar 28, 2026• 1 min read
What it is
A train/test split separates data into one portion for learning patterns and another portion for checking whether the model generalizes.
Why it matters
Without a held-out test set, a model may look impressive only because it memorized the training examples.
Example
If you have 1,000 rows, you might train on 800 rows and test on 200 rows.
Common mistakes
- evaluating on the same data used for training
- doing the split after leakage has already happened
- forgetting stratification when the classes are imbalanced
Key takeaway
A good score is only meaningful if it comes from data the model did not see during training.