Homework Help is Here – Start Your Trial Now!
Engineering
Data Structures and Algorithms
1. What further questions would you ask on the evaluation? Think of test data, metrics, and baselines. 2. What would be potential privacy risks related to re-identification or the revelation of sensitive information of customers to the data science team? How to measure these? 3. Might there be discrimination against sensitive groups, such as Muslims or women, if the payment data is used? How to evaluate? Might there be certain features (account numbers) that if a customer made a payment to those, the sensitive attribute is revealed? How to measure whether the model is using these in a discriminatory way? 4. Would the invitees of the golf tournament event require an explanation for their predicted interest? If so, what type of explanation would you provide?

1. What further questions would you ask on the evaluation? Think of test data, metrics, and baselines. 2. What would be potential privacy risks related to re-identification or the revelation of sensitive information of customers to the data science team? How to measure these? 3. Might there be discrimination against sensitive groups, such as Muslims or women, if the payment data is used? How to evaluate? Might there be certain features (account numbers) that if a customer made a payment to those, the sensitive attribute is revealed? How to measure whether the model is using these in a discriminatory way? 4. Would the invitees of the golf tournament event require an explanation for their predicted interest? If so, what type of explanation would you provide?

icon
Related questions
Question
+ ill You're the ChiefData Science Officer at a large bank. You've instructed your team to experiment with using payment data for marketing purposes, predicting which customer might be interested in a golf tournament that the bank sponsors. So the data instances correspond to customers, and the features are unique account numbers. Your newly hired team is ready to shine and has put quite some effort in building a linear model, where each ac-count number that one can pay to is given a coefficient. The prediction model hence predicts interest based on whom the customer has made payments to. They proudly report to you that the accuracy of their model is 95%, on a test set chosen in January. 1. What further questions would you ask on the evaluation? Think of test data, metrics, and baselines. 2. What would be potential privacy risks related to re-identification or the revelation of sensitive information of customers to the data science team? How to measure these? 3. Might there be discrimination against sensitive groups, such as Muslims or women, if the payment data is used? How to evaluate? Might there be certain features (account numbers) that if a customer made a payment to those, the sensitive attribute is revealed? How to measure whether the model is using these in a discriminatory way? 4. Would the invitees of the golf tournament event require an explanation for their predicted interest? If so, what type of explanation would you provide? M Mint un pur bat as su
Transcribed Image Text:+ ill You're the ChiefData Science Officer at a large bank. You've instructed your team to experiment with using payment data for marketing purposes, predicting which customer might be interested in a golf tournament that the bank sponsors. So the data instances correspond to customers, and the features are unique account numbers. Your newly hired team is ready to shine and has put quite some effort in building a linear model, where each ac-count number that one can pay to is given a coefficient. The prediction model hence predicts interest based on whom the customer has made payments to. They proudly report to you that the accuracy of their model is 95%, on a test set chosen in January. 1. What further questions would you ask on the evaluation? Think of test data, metrics, and baselines. 2. What would be potential privacy risks related to re-identification or the revelation of sensitive information of customers to the data science team? How to measure these? 3. Might there be discrimination against sensitive groups, such as Muslims or women, if the payment data is used? How to evaluate? Might there be certain features (account numbers) that if a customer made a payment to those, the sensitive attribute is revealed? How to measure whether the model is using these in a discriminatory way? 4. Would the invitees of the golf tournament event require an explanation for their predicted interest? If so, what type of explanation would you provide? M Mint un pur bat as su
Expert Solution
steps

Step by step

Solved in 6 steps

SEE SOLUTIONCheck out a sample Q&A here
Blurred answer