Short Definition
A train/test split separates data into training and evaluation sets to measure generalization.
Definition
A train/test split is a data partitioning strategy in which a dataset is divided into at least two subsets: one used to train the model and one used to evaluate its performance. The test set is not used during training and serves as an approximation of unseen data.
The goal is to estimate how well the model generalizes beyond the data it was trained on.
Why It Matters
Evaluating a model on the same data used for training leads to overly optimistic performance estimates. A proper train/test split helps detect overfitting and provides a more realistic assessment of model performance.
It is a foundational practice in machine learning experimentation.
How It Works (Conceptually)
- The dataset is randomly or systematically divided
- The model is trained only on the training set
- Performance is measured on the test set
- Results approximate real-world behavior
The test set acts as a proxy for unseen data.
Minimal Python Example
train_data, test_data = split(dataset, test_ratio=0.2)Common Pitfalls
- Data leakage between train and test sets
- Tuning hyperparameters on the test set
- Using inappropriate splits for time-series data
- Assuming a single split is always sufficient
Related Concepts
- Generalization
- Cross-Validation
- Data Leakage
- Evaluation Metrics
Neural Network Lexicon 