Test legat de ghid · Mediu Nivel
AI în Retail Quiz
Evaluați IA cu amănuntul în funcție de contextul deciziei, impactul asupra consumatorilor, rezultatele subgrupului, confidențialitatea, căile de corecție și costul total al deciziilor false.
Întrebare 1 de 8
Why define data available at the decision moment for a retail model?
Previzualizează fiecare întrebare8 întrebări
Răspunsurile și explicațiile apar pe măsură ce susțineți testul.
- Why define data available at the decision moment for a retail model?
- Recommendations, pricing, forecasts, and fraud reviews have different inputs and error costs
- Every decision can use later purchase data
- Data timing never matters
- All retail models optimize the same outcome
- A recommender increases basket size but customers complain about irrelevant offers. What should be reviewed?
- Only the basket-size increase
- Customer outcomes and whether the proxy metric harms service
- The color of the recommendation card
- The number of products in inventory
- What did the false-decline worked example ask retailers to measure?
- The average basket size
- Only the number of blocked fraud attempts
- Legitimate customers declined, correction time, and safer verification options
- The model’s response time
- Why evaluate retail outcomes by subgroup?
- It removes the need to monitor returns
- All customers have identical histories
- Segmentation guarantees better revenue
- Averages can hide less useful service or unfair treatment for some groups
- Which outcomes should accompany a recommendation model’s sales results?
- Only click-through
- Useful discovery, returns, complaints, and stock availability
- Only the count of messages
- Numărul de parametri ai modelului
- Which retail information does the guide identify as needing protection?
- Only store hours
- Only public product names
- Purchase history, location, and identity information
- The model’s display color
- What should an automated system provide when it cannot answer or a transaction is wrong?
- An alternative route and a way for staff to correct the issue
- A higher price
- A hidden score only
- No explanation or recourse
- Why monitor new products, promotions, and seasons?
- They prevent the need for versioning
- They make all recommendations more accurate
- They eliminate false declines
- They can change behavior and cause model or policy drift