# Predictive Analytics Quiz

## Predictive Analytics Quiz >> Customer Analytics

1. *Please note: Some of the questions you’ll find in this quiz aren’t just straight “regurgitation” of the materials presented in the videos – and this is exactly what I had in mind. These questions will force you to think a bit about the concepts/methods covered, but careful consideration of the course material should lead you to an indisputable correct answer in every case.Here’s one hint: make sure you look at the presentation decks as well as the videos.***In which of these situations would it be more appropriate to use a probability model rather than a regression/data-mining approach?**

- Predicting whether the customer will buy the brand at least once in the next year
- Predicting the brand that the customer will buy during her next category purchase
**Predicting when the customer will make her next purchase✅**- Predicting which customer is most likely to churn in the next year
- Predicting whether the customer will churn in the next year

2. *Please note: Some of the questions you’ll find in this quiz aren’t just straight “regurgitation” of the materials presented in the videos – and this is exactly what I had in mind. These questions will force you to think a bit about the concepts/methods covered, but careful consideration of the course material should lead you to an indisputable correct answer in every case.**Here’s one hint: make sure you look at the presentation decks as well as the videos.***Which of the following are genuine data-mining procedures? (Please check all that apply)**

- All answers are correct
- SCAN
**MARS✅****CART✅**

3. *Please note: Some of the questions you’ll find in this quiz aren’t just straight “regurgitation” of the materials presented in the videos – and this is exactly what I had in mind. These questions will force you to think a bit about the concepts/methods covered, but careful consideration of the course material should lead you to an indisputable correct answer in every case.**Here’s one hint: make sure you look at the presentation decks as well as the videos.***Which of these statements is most aligned with our assumption(s) about randomness when it comes to modeling/explaining customer behavior?**

- Most customers are predictable but there is usually a segment of “as if” random ones that should be accounted for
- Any given customer is quite predictable, but the randomness exists across customers
- We make some assumptions about randomness in order to derive the mathematical model, but when it comes to actually estimating the model they no longer apply
**Customers are not truly random but appear to be “as if” random from an outsider observer’s perspective✅**- Each customer is assumed to behave randomly in accordance with a standard normal (“bell-shaped”) distribution

4. *Please note: Some of the questions you’ll find in this quiz aren’t just straight “regurgitation” of the materials presented in the videos – and this is exactly what I had in mind. These questions will force you to think a bit about the concepts/methods covered, but careful consideration of the course material should lead you to an indisputable correct answer in every case.**Here’s one hint: make sure you look at the presentation decks as well as the videos.***Among the explanations below, which one is a reason to favor a probability model over a regression-like (e.g., data-mining) model for long-run projections of customer behavior?**

- Probability models are more accurate than regression models
- Probability models can determine customer motivations
- If the observed behavior is viewed in an “as if” random manner, it would be wrong to put it into a regression-like model as if it’s deterministically true
**Regression-like models are fine for a one-period-ahead prediction, but not beyond that horizon✅**

5. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***Why does the “RFM” rubric present the three key measures (recency, frequency, and monetary value) in that order?**

- This is the order in which they were discovered/identified as being highly predictive of future behavior
- Recency is the easiest of the three to observe/measure
**Recency is the most predictive of the three✅**- Recency and frequency are equally important, and monetary value is far less important than both of them
- There is no particular reason; it’s just an arbitrary order

6. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***When we refer to a “cohort,” we are talking about a group of customers who:**

**Share similar acquisition characteristics (e.g., time of acquisition)✅**- Share similar purchasing propensities
- Share similar observable personal characteristics (e.g., demographics)
- Share similar responsiveness to marketing tactics
- Share similar churn propensities

7. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***Referring back to the dataset (and model) we covered extensively, how would these two customers (both “acquired” in 1995) compare to each other, in terms of their expected future purchasing?**

1995 1996 1997 1998 1999 2000 2001 2002Vrinda 1 0 0 0 1 1 1 0Yoshinori 1 1 1 1 0 0 1 0

- Vrinda would likely be more valuable
- There’s not enough information here to make the decision
- They would be expected to be roughly equal
**Yoshinori would likely be more valuable✅**

8. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***What does a “BTYD” model refer to?**

- Buy Till You Die
**Bayesian Transformation of Yearly Data✅**- Beta Time-Yield Distribution
- Back-Test Your Data

9. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***Referring back to the dataset (and model) we covered extensively, how would these two customers (both “acquired” in 1995) compare to each other, in terms of their expected future purchasing?**

1995 1996 1997 1998 1999 2000 2001 2002Ted 1 0 0 0 1 0 1 0Jane 1 1 1 1 0 0 0 0

**Jane would likely be more valuable✅**- Ted would likely be more valuable
- There’s not enough information here to make the decision
- They would be expected to be roughly equal

10. *Here’s one hint: make sure you look at the presentation decks as well as the videos.***Which of these real actions would not be represented by the “buy” in the BTYD model?**

- When a customer files an insurance claim
- When a customer renews a subscription
- When a customer participates in a promotional sale
- When a customer attends a sales event
**All answers are possibilities✅**