Question

What is a common reason for an ml model that works well in training but fails in production coursera

Answer 1

Training-serving skew

Step-by-step explanation:

Training-serving skew is a term in Machine Learning that is used to describe the discrepancy between performance in training and performance in real serving. The cause of the Training-serving skew can be any of the following:

1. Disparity in the handling of data between training compared to serving

2. Difference in data used in training and serving

3. Feedback loop between model and algorithm used.

Hence, the common reason for an ml model that works well in training but fails in production is called TRAINING - SERVING SKEW