Question-11: You are working in retail bank as a business analyst, this bank offers the various retail products like Commercial Loan, Mortgage Loan, Credit Card, saving accounts, Current account etc. as regular retail bank does. Your bank has done collaboration with one of the eCommerce company to share their customer information with the bank as well as providing discounts if customer use the credit card issued by a bank. There is around 10GB of the data on daily basis in the evening you are receiving as well as in the real-time also you are receiving user information as soon as he does the transactions. You have been assigned with the task to identify below two things

• Whether customer can buy a new product offered by bank like loan, credit card tops up etc. and prediction should be Yes or no only.
• Similarly, at the real time the information you are receiving about the user transaction you need to find whether this is a fraud transaction or not.

Which of the following you would be using to solve the given analytical problem?

1. You would be using Apache Spark and EMR to run your Machine Learning algorithm in real time.
2. Get the Latest AWS Certification Questions & Answer based on recent syllabus from this link
3. You would be considering to use the Binary Classification Model
4. You would be considering Multiclass Machine Learning Model because of more than two products offered by bank.
5. Get the Latest AWS Certification Questions & Answer based on recent syllabus from this link
6. You would be using batch predictions for asynchronously predicting the fraud transactions for the data you are continuously getting.

Ans: B, C, E

Exp: You need to understand Get the latest AWS Training, Certification Preparation Material, Books & Interview questions  the specific thing which is being asked in this question. Read carefully and summarize the things.

• Bank offering multiple products but they wanted to predict whether user/customer will subscribe or not the product like Yes or No
• To find whether the transaction is fraud or not, again need to predict either of the two values.

Hence, best fit for this requirement is a binary classification (where you need to predict either of the one value). Hence 3^rd option is correct.

Next, we are received data in both batch as well as in real-time and AWS ML supports predictions on both kind of the data. If data volume is less than 100GB.

Hence, we can use Batch predictions in the evening for asynchronously predicting the values. And if we want to predict the values in real-time it means synchronous prediction is required. Hence, we would be using the AWS ML Real-time predictions. Hence, in this case option-2 and option-5 are also fit well.

Option-4 is given to confuse with the type of the products offered by the bank and that is wrong answer.

Option-1 is a possible solution, but why do you want to create such a complex solution even it is possible with the simple setup using AWS ML.