Databricks Machine Learning Associate Question 617
Single answerA team is training a machine learning model on a large dataset using Databricks. Initially, they use a simple linear regression model, but the predictions are highly inaccurate. They then switch to a more complex deep learning model, which performs well on the training data but poorly on the validation dataset. What is the most likely reason for the poor validation performance of the deep learning model?
- A
The deep learning model has high bias, leading to underfitting.
- B
The deep learning model has high variance, leading to overfitting.
- C
The deep learning model is not complex enough for the dataset.
- D
The linear regression model has overfitted the data, affecting the deep learning model's performance.
Show answer and explanation
Correct answer: B
Explanation
The poor validation performance of the deep learning model is due to overfitting, a common issue with highly complex models. Overfitting occurs when a model captures noise or patterns specific to the training data, resulting in high variance. This leads to poor generalization on unseen data, as observed in the validation set.
- A. Incorrect.
High bias is typically associated with simpler models that fail to capture the complexity of the data, leading to underfitting. This does not explain the deep learning model's poor validation performance.
- B. Correct.
High variance occurs when a model is overly complex and fits the training data too closely, failing to generalize well to unseen data. This explains why the deep learning model performs poorly on the validation dataset.
- C. Incorrect.
The deep learning model is already very complex. A lack of complexity would not result in poor validation performance in this case.
- D. Incorrect.
The performance of the linear regression model is unrelated to the deep learning model's validation performance, as each model is trained independently.