P2.T9.803. Machine learning (van Liebergen)

Nicole Seaman

Director of CFA & FRM Operations
Staff member
Subscriber
Learning objectives: Describe the process of machine learning and compare machine learning approaches. Describe the application of machine learning approaches within the financial services sector and the types of problems to which they can be applied. Analyze the application of machine learning in three use cases: Credit risk and revenue modeling; Fraud; Surveillance of conduct and market abuse in trading.

Questions:

803.1. Peter the risk analyst is helping his international financial services client analyze their very big client transaction database. His immediate task is to conduct an anti-money laundering (AML) analysis. Unlike credit card fraud, however, money laundering is hard to define: there is no universally agreed definition of money laundering. Consequently the historical database contains no field indicating whether a transaction was fraudulent or not; put another way, there is no dependent variable. As such, Peter effectively only has input variables with which to work. If his goal is to yield insights from the data for his client, which of the following methods (among the choices given) in this situation is the MOST appropriate?

a. Clustering
b. Support vector machines
c. Classification decision tree
d. LASSO, a penalized regression


803.2. In regard to machine learning, each of the following statements is true EXCEPT which is inaccurate?

a. Averaging over many small models tends to give better out-of-sample prediction than choosing a single model
b. Deep learning is a supervised learning method that requires structured data but the layers of features are designed by human engineers
c. There is often a trade-off between prediction and explanation and many machine learning methods are better at prediction than explanation
d. Over-fitting is a common problem in non-parametric, non-linear machine learning models, and its symptoms include very good in-sample fit but poor out-of-sample performace


803.3. Among these choices, which of the following machine learning models is the LEAST useful for the regulatory purpose of providing a system that can be audited and verified by the supervisor?

a. Logit regression
b. Linear regression approach
c. Machine learning ensemble
d. Behavioral science-based model

Answers here:
 
Last edited by a moderator:
Top