StackedML
Practice
Labs
Questions
Models
Pricing
Sign in
Questions
/
Model Evaluation & Experimentation
/
Validation
/
Train/test split
← Previous
Next →
92.
Choosing Train/Test Split Ratio
easy
A common train/test split ratio is 80/20. What factor most influences the choice of split ratio?
A
Class imbalance — the split ratio should match the minority class proportion in the dataset
B
Model complexity — more complex models require larger test sets to detect overfitting reliably
C
Dataset size — larger datasets can afford a smaller test fraction while still providing reliable evaluation
D
Feature dimensionality — high-dimensional data requires more test samples to cover the feature space
Sign in to verify your answer
← Back to Questions