Train/Test Dataset SVM Classifier Preprocessing K-Means...

Thanks for your feedback!

A bank is trying to implement a fraud detection algorithm in order to identify fraudulent transactions. A typical flowchart of the fraud detection process is given below: The figure shows 6 major steps and a total of 700 sub-steps (not shown) across all 6 major steps. Out of these sub-steps, 525 can be executed in parallel using the bank's existing machines, while the remaining sub-steps can only be executed sequentially. Given this information, please answer the following questions: 1. Calculate the Speedup and Efficiency to execute the fraud detection process on a cluster with n = 256 nodes (assuming the workload is not scaled up and regular parallel processing occurs). [10 marks] 2. What are your comments on the efficiency? Is it low or high? 3. What could be the possible reason for the low/high efficiency from a maximum Speedup point of view? [5 marks] 4. If the fraud detection process is not run on a distributed setup, the firm is considering two options: using VMs or containers. Recommend and justify which one of the two options you think the bank should opt for? [5 marks] Train/Test Dataset SVM Classifier Applying SVM on test data Fraud Detection Process K-Means Clustering Fraud (1) or not Fraud (0) Feature KNN

Play audio

Feedback

Question

Train/Test Dataset SVM Classifier Preprocessing K-Means Clustering Feature Selection using KNN Applying SVM on test data Fraud (1) or not Fraud (0) Fraud Detection Process

Question

Train/Test Dataset SVM Classifier Preprocessing K-Means Clustering Feature Selection using KNN Applying SVM on test data Fraud (1) or not Fraud (0) Fraud Detection Process

Please give Ace some feedback