The aim of this industrial thesis topic is to design a system that rates the possibility of an online seller of a product from conducting fraudulent transactions i.e. accepting the money from a buyer but not providing the desired product. A database of past online transactions that are classified as either good or bad will be used as input to the system. The resultant 'seller risk score' may be used as a part of an automated system that assesses online transactions to determine whether they are likely to be good, or have a high chance of being bad and need to be manually reviewed. A key concern is finding a good balance between false positives and false negatives, as this will mean either a liability to the payment company or a loss of business opportunity and customer dissatisfaction.
Comments:
Would suit a student with a strong background in machine learning as well as finance.
