Question
A company is training an AI model using a custom table to determine the amount of time it takes to deliver a package based on several key fields.
The testing data used to train the model is used for all training and regression testing scenarios and is considered complete data.
The trained model predicts a 2 percent variance between the estimated delivery time and the actual delivery time of packages.
The executive sponsors reject the model because the actual variance is at 15 percent.
You need to address the sponsors' concern.
What should you do?
- A. Replace the training data with real-world data. Correct
- B. Reduce the size of the data used within the model.
- C. Increase the size of the data used with the model.
- D. Use sample training data from Microsoft.
Explanation
Replacing the training data with real-world data can improve model accuracy by better reflecting actual conditions. Reducing or increasing data size (options B and C) without addressing data quality issues may not resolve variance problems.
