Describe Core Machine Learning Concepts AI-900 Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A university is building a model to predict student graduation. They have 10 years of student records, including grades, attendance, and graduation status. The team splits the data into training and validation sets. What is the primary purpose of the training set?

   • ATo evaluate the final model's accuracy
   • BTo replace the validation set
   • CTo tune hyperparameters
   • DTo teach the model the relationship between features and the label
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A marketing team has a dataset of 100,000 customer records with a binary column indicating whether each customer responded to a past campaign. They want to train a model to predict future responses. The dataset is ordered by the date the record was collected. Which splitting strategy is most appropriate?

   • ARandomly split the data into 80% training and 20% validation
   • BUse the first 20% for training and the last 80% for validation
   • CDo not split; train on all data and evaluate on the same data
   • DSplit chronologically so the training set contains earlier records and the validation set contains later records
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A company has a dataset of 5,000 customer transactions with a label indicating fraud. They want to train a model and evaluate it. Which approach is most appropriate?

   • ASplit the data randomly into 80% training and 20% validation.
   • BUse the first 4,000 records for training and the last 1,000 for validation.
   • CUse 50% for training and 50% for validation, but only if the data is sorted by transaction amount.
   • DTrain on all 5,000 records and validate on a random 20% of the same records.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   What is the primary purpose of a training dataset in machine learning?

   • ATo evaluate the final model's performance on unseen data
   • BTo store the model's predictions for future reference
   • CTo define the features and labels that will be used in the model
   • DTo provide examples that the model uses to learn patterns and adjust its parameters
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   An e-commerce company wants to predict whether a customer will return a product. They have historical data with columns: CustomerAge, ProductCategory, OrderTotal, and Returned (Yes/No). What should be the label in this supervised learning scenario?

   • AThe OrderTotal column
   • BThe Returned column (Yes/No)
   • CThe ProductCategory column
   • DThe CustomerAge column
   Reveal Solution Discussion Select an answer first