DataTerminology.comNEW
Also known as: High Bias, Insufficient Complexity
When a model is too simple to capture patterns, producing high error on both training and test data.
DESCRIPTION
Underfitting reflects high bias—models cannot represent the underlying relationship. Indicators include high training error and poor validation performance. Causes include overly simple hypotheses, excessive regularization, or inadequate features. Fixes include increasing model capacity, reducing regularization, enriching features, and revisiting problem framing. Monitoring learning curves helps distinguish underfitting from overfitting.
ALTERNATIVE DEFINITIONS
Wikipedia
Underfitting occurs when a statistical model or machine learning algorithm cannot capture the underlying trend of the data.
Defines underfitting as failure to capture underlying trends due to overly simple models.
Google Developers
Underfitting happens when a model is too simple to learn the underlying pattern of the data, resulting in poor performance.
Stresses insufficient model complexity as root cause.
RELATED TERMS
Bias-Variance Tradeoff Model Capacity Feature Engineering Regularization Training Error Polynomial Degree K-Nearest Neighbors Linear Model Hyperparameter Tuning Learning Curve
COMPETING TERMS
RELATIONS
- Broader: Underfitting
KNOWLEDGE AREAS
TARGET AUDIENCE
Data Scientist AI Engineer MLOps Engineer
EXTERNAL SOURCES
- Wikipedia - Overfitting
- Google Developers - Explanation of representation, underfitting, overfitting
- scikit-learn - Using learning curves to diagnose fit
TOP QUESTIONS
How do you identify Model Underfitting?
High error on both training and validation sets indicates underfitting. Learning curves that plateau at high error reinforce the diagnosis.
Category: Testing
What causes Model Underfitting?
Overly simple models, too much regularization, or insufficient features prevent capturing complex relationships.
Category: Overview
How do you fix Model Underfitting?
Increase model complexity, reduce regularization, engineer richer features, or collect more informative data.
Category: Best practices
Is Underfitting preferable to Overfitting?
Neither is ideal. In practice, slight overfitting may be tolerable if generalization is adequate. Aim for balanced bias and variance.
Category: Opinion
What is the role of feature engineering in Underfitting?
Adding relevant features or transformations can increase model capacity to capture relationships without overcomplicating architecture.
Category: Management
How does regularization affect Underfitting?
Excessive regularization shrinks coefficients and can oversimplify the model, increasing bias and underfitting.
Category: Explanation
Can more data solve Underfitting?
More data helps if the model has enough capacity. Otherwise, more data alone may not resolve high bias.
Category: Misunderstanding
Which metrics reveal Underfitting?
Watch training loss and validation loss; both staying high indicates underfitting. Simple baselines outperforming your model is another sign.
Category: Testing
What’s a hot topic related to Underfitting?
Pretraining and transfer learning reduce data requirements and help simple architectures capture richer patterns.
Category: Topic