Question

Leggere, an internet book retailer, is interested in better understanding the purchase decisions of its customers. For a set of 2,000 customer transactions, it has categorized the individual book purchases comprising those transactions into one or more of the following categories: Novels, Willa Bean series, Cooking Books, Bob Villa Do-It-Yourself, Youth Fantasy, Art Books, Biography, Cooking Books by Massimo Bottura, Harry Potter series, Florence Art Books, and Titian Art Books. Leggere has conducted association rules analysis on this dataset and would like to analyze the output. Based on a minimum support of 200 transactions and a minimum confidence of 50%, the table below shows the top 10 rules with respect to lift ratio. (a) Explain why the top rule "If a customer buys a Bottura cooking book, then they buy a cooking book," is not helpful even though it has the largest lift and 100% confidence. A Bottura cooking book is a specific type of cooking book, so whenever a customer purchases a Bottura cooking book they must also have bought a cooking book. Thus, the confidence of "If a customer purchases a Bottura cooking book, then they purchase a cooking book" is 100%. Thus, any rule with the antecedent being a subset of the consequent will have a confidence of 100% and provides no information. Furthermore, if we would interpret this rule as suggesting that if a customer is interested in a specific type of cooking book that they may be interested in other types of cooking books, this still provides no profound insight. (b) Explain how the confidence of 52.99% and lift ratio of 2.20 was computed for the rule "If a customer buys a cooking book and a biography book, then they buy an art book." Interpret these quantities. A confidence of 52.99% means that for 52.99% of the transactions when a cooking book and biography are purchased, an art book is purchased. A lift ratio of 2.20 means that a transaction in which a cooking book and biography is purchased is 120% more likely to also have purchased an art book than a randomly-selected transaction. (c) Based on these top 10 rules, what general insight can Leggere gain on the purchase habits of these customers? In the top 10 rules, there are two rules involving the item set {cooking, Bob Villa, art} and two rules involving the item set {biography, novels, cooking}. Leggere may be well served to investigate promotions involving these respective item sets. (d) What will be the effect on the rules generated if Leggere decreases the minimum support and reruns the association rules analysis? Decreasing the minimum support will generate a larger pool of rules and therefore will likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, the transactions involving the item set of the association rule may not occur as frequently due to the lowered minimum support. Thus, the association rule may be spurious, or the item set may apply to too few transactions to be a valuable business opportunity. (e) What will be the effect on the rules generated if Leggere decreases the minimum confidence and reruns the association rules analysis? Decreasing the minimum confidence will generate a larger pool of rules and therefore likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, if this rule has a low confidence then this means that there also is a low probability that the consequent will occur given the antecedent. That is, as the confidence of an association rule decreases, the relationship between the antecedent and consequent will fail more often. Depending on the profit margin associated with the transaction, this may make the association rule not a viable business opportunity.

Name: leggere an internet book retailer is interested in better understanding the purchase decisions of its customers for a set of 2000 customer transactions it has categorized the individual book 45443
Uploaded: 2024-05-13T17:16:21-08:00
Duration: 7 min 31 s
Channel: Shu Naito
Description: leggere an internet book retailer is interested in better understanding the purchase decisions of its customers for a set of 2000 customer transactions it has categorized the individual book 45443

(a) Explain why the top rule "If a customer buys a Bottura cooking book, then they buy a cooking book," is not helpful even though it has the largest lift and 100% confidence.

A Bottura cooking book is a specific type of cooking book, so whenever a customer purchases a Bottura cooking book they must also have bought a cooking book. Thus, the confidence of "If a customer purchases a Bottura cooking book, then they purchase a cooking book" is 100%.

Thus, any rule with the antecedent being a subset of the consequent will have a confidence of 100% and provides no information. Furthermore, if we would interpret this rule as suggesting that if a customer is interested in a specific type of cooking book that they may be interested in other types of cooking books, this still provides no profound insight.

(b) Explain how the confidence of 52.99% and lift ratio of 2.20 was computed for the rule "If a customer buys a cooking book and a biography book, then they buy an art book." Interpret these quantities.

A confidence of 52.99% means that for 52.99% of the transactions when a cooking book and biography are purchased, an art book is purchased.

A lift ratio of 2.20 means that a transaction in which a cooking book and biography is purchased is 120% more likely to also have purchased an art book than a randomly-selected transaction.

In the top 10 rules, there are two rules involving the item set {cooking, Bob Villa, art} and two rules involving the item set {biography, novels, cooking}. Leggere may be well served to investigate promotions involving these respective item sets.

(d) What will be the effect on the rules generated if Leggere decreases the minimum support and reruns the association rules analysis?

Decreasing the minimum support will generate a larger pool of rules and therefore will likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, the transactions involving the item set of the association rule may not occur as frequently due to the lowered minimum support. Thus, the association rule may be spurious, or the item set may apply to too few transactions to be a valuable business opportunity.

(e) What will be the effect on the rules generated if Leggere decreases the minimum confidence and reruns the association rules analysis?

Decreasing the minimum confidence will generate a larger pool of rules and therefore likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, if this rule has a low confidence then this means that there also is a low probability that the consequent will occur given the antecedent. That is, as the confidence of an association rule decreases, the relationship between the antecedent and consequent will fail more often. Depending on the profit margin associated with the transaction, this may make the association rule not a viable business opportunity.

Added by Anita D.

Elementary Statistics a Step by Step Approach

Allan G. Bluman 9th Edition

Instant Answer

Solved by Expert Shu Naito

05/13/2024

Step 1

Step 1: The top rule "If Bottura cooking book => cooking book" is uninformative because Bottura cooking books are a subtype of cooking books, so the rule is tautological; antecedent ⊆ consequent yields confidence = 100% but no actionable insight. Show more…

Show all steps

Thanks for your feedback!

Leggere, an internet book retailer, is interested in better understanding the purchase decisions of its customers. For a set of 2,000 customer transactions, it has categorized the individual book purchases comprising those transactions into one or more of the following categories: Novels, Willa Bean series, Cooking Books, Bob Villa Do-It-Yourself, Youth Fantasy, Art Books, Biography, Cooking Books by Massimo Bottura, Harry Potter series, Florence Art Books, and Titian Art Books. Leggere has conducted association rules analysis on this dataset and would like to analyze the output. Based on a minimum support of 200 transactions and a minimum confidence of 50%, the table below shows the top 10 rules with respect to lift ratio. (a) Explain why the top rule "If a customer buys a Bottura cooking book, then they buy a cooking book," is not helpful even though it has the largest lift and 100% confidence. A Bottura cooking book is a specific type of cooking book, so whenever a customer purchases a Bottura cooking book they must also have bought a cooking book. Thus, the confidence of "If a customer purchases a Bottura cooking book, then they purchase a cooking book" is 100%. Thus, any rule with the antecedent being a subset of the consequent will have a confidence of 100% and provides no information. Furthermore, if we would interpret this rule as suggesting that if a customer is interested in a specific type of cooking book that they may be interested in other types of cooking books, this still provides no profound insight. (b) Explain how the confidence of 52.99% and lift ratio of 2.20 was computed for the rule "If a customer buys a cooking book and a biography book, then they buy an art book." Interpret these quantities. A confidence of 52.99% means that for 52.99% of the transactions when a cooking book and biography are purchased, an art book is purchased. A lift ratio of 2.20 means that a transaction in which a cooking book and biography is purchased is 120% more likely to also have purchased an art book than a randomly-selected transaction. (c) Based on these top 10 rules, what general insight can Leggere gain on the purchase habits of these customers? In the top 10 rules, there are two rules involving the item set {cooking, Bob Villa, art} and two rules involving the item set {biography, novels, cooking}. Leggere may be well served to investigate promotions involving these respective item sets. (d) What will be the effect on the rules generated if Leggere decreases the minimum support and reruns the association rules analysis? Decreasing the minimum support will generate a larger pool of rules and therefore will likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, the transactions involving the item set of the association rule may not occur as frequently due to the lowered minimum support. Thus, the association rule may be spurious, or the item set may apply to too few transactions to be a valuable business opportunity. (e) What will be the effect on the rules generated if Leggere decreases the minimum confidence and reruns the association rules analysis? Decreasing the minimum confidence will generate a larger pool of rules and therefore likely result in association rules with larger lift ratios. A rule with a larger lift ratio suggests that it is effective at identifying transactions in which the consequent occurs. However, if this rule has a low confidence then this means that there also is a low probability that the consequent will occur given the antecedent. That is, as the confidence of an association rule decreases, the relationship between the antecedent and consequent will fail more often. Depending on the profit margin associated with the transaction, this may make the association rule not a viable business opportunity.