CLOUD-DIGITAL-LEADER · Question #3
Your team has developed a machine learning model for your customer. The test results indicate very strong predictive capability. The model is then deployed in production. Evaluation of the…
The correct answer is D. The model is overfitted. Train with more data. This is a classic overfitting scenario. The model memorized the training/test data too precisely (including its noise and idiosyncrasies) rather than learning generalizable patterns. As a result, it performs well on seen data but poorly on unseen production data. The standard…
Question
Your team has developed a machine learning model for your customer. The test results indicate very strong predictive capability. The model is then deployed in production. Evaluation of the predictions in production show that they are off by a pronounced margin. What is the issue and how can you solve for it?
Options
- AThe model is under fitted. Train with less data.
- BThe model is over fitted. Add more features to the model to fix it.
- CThe model is fine since the test results are good. Fix the production of incoming data.
- DThe model is overfitted. Train with more data.
How the community answered
(48 responses)- A8% (4)
- B13% (6)
- C2% (1)
- D77% (37)
Explanation
This is a classic overfitting scenario. The model memorized the training/test data too precisely (including its noise and idiosyncrasies) rather than learning generalizable patterns. As a result, it performs well on seen data but poorly on unseen production data. The standard solution is to train with more diverse data, which forces the model to learn broader patterns rather than memorizing specific examples. Underfitting (A) would show poor test results too, not just production results. Adding more features (B) typically worsens overfitting. Blaming production data (C) ignores the root cause.
Topics
Community Discussion
No community discussion yet for this question.