Data Drift — Learn free

Self-check

Questions 1–10 of 10

1. In which practice is the concept of Data Drift used?
Question 1A. Project managementB. Data ScienceC. Network securityD. ProcurementE. Master Data Management
2. What is Data Drift?
Question 2A. A change in the data elements a model usesB. A change of data within the same selected data elementsC. A change in the model's codeD. A change in the project scopeE. A mismatch between two reports
3. By which other names is Data Drift known?
Question 3A. Covariate shift, virtual drift or virtual concept driftB. Class drift, real concept drift or posterior probability shiftC. Sudden drift, gradual drift or incremental driftD. Scope creep or requirement driftE. Reconciliation drift or discrepancy drift
4. In the products example, what causes the Data Drift?
Question 4A. Salary data replacing product dataB. The model's logic being rewrittenC. The list of product types growing from 10 during training to 15 in productionD. The weather changing unexpectedlyE. The database being migrated to the cloud
5. In the products example, how might model accuracy change after Data Drift?
Question 5A. It stays at 85%B. It rises to 95%C. It drops from around 85% to perhaps below 70%D. It becomes exactly zeroE. It cannot be measured
6. Which statement about Data Drift is correct?
Question 6A. It means there is something wrong with the modelB. It means the model's data elements have changedC. It is a completely unexpected behaviourD. It is one of the most expected behaviours in every data modellingE. It only affects OLTP systems
7. Why must an organisation have automated ways to track and identify Data Drift?
Question 7A. Because regulators require itB. Because drift changes the project scopeC. Because drift corrupts the databaseD. Because drift increases cloud costsE. Because after drift there is a high chance the model will not predict with the planned accuracy, and wrong predictions mean wrong decisions
8. What is the remedy for Data Drift?
Question 8A. Retrain the model quickly in the training environment and deploy it againB. Change the data elements used by the modelC. Delete the new dataD. Switch from SQL to MDXE. Add more processors
9. What does 'drift' mean in general?
Question 9A. Summarising data into totalsB. Deleting data after retention expiresC. Copying data between nodesD. Shifting or moving from one point to another, but not in a standard approachE. Encrypting data in transit
10. Which scenario would NOT count as Data Drift in the products example?
Question 10A. Five new product types appearing in productionB. Product data extending beyond the training setC. Product values the model was not trained onD. The model's accuracy falling because of unrecognised productsE. Salary data starting to arrive instead of product data