Both the training error and the validation error are high and close to each other. What does that suggest?
Options
- High variance: the model is overfitting
- High bias: the model is too simple for the pattern (underfitting) (Correct option)
- Data leakage from the validation set
- A validation set that is too small
Answer
High bias: the model is too simple for the pattern (underfitting)
Why
When a model is bad even on the data it trained on, it cannot capture the pattern: that is high bias, or underfitting.
Overfitting (high variance) looks different: low training error but much higher validation error.
Fixes for high bias: a more flexible model, better features, less regularization. More data alone will not help much.
Remember
Both errors high: bias. A big gap between them: variance.