Sunteți pe pagina 1din 4

1.

SAS test
2. what are the major variables you plan to use if you want to build a model to predict
customer charge off risk?
3. Credit Product knowledge
4. how to validate a model in the sample or out of the sample
5. technical questions around modeling time serires,
6. assumptions of OLS, methods of detecting outliers given data.
7. Maximum Likelihood Estimation, p-values, variable selection, Monte-Carlo Simulations,
Mean Value Theory and Stochastic Calculus
8. For example, if you are interviewing for a role with in model validation, don’t be surprised
if technical model development or regulatory knowledge-based questions arise
9. What are the main steps in validation?

Domain Specific
 What are the differences between Mortgage, Personal Loan and Credit Card
products?
 What are the key changes happened from Basel I to Basel II?
 What are the 3 pillars of Basel II?
 What is Vintage Analysis?
 What to you understand by reject inference? Why do you need? What are the
commonly used methods for reject inference?
 What is score alignment? How do you correct?
 What is the difference between bankcard, Private label credit cards and co-brand
cards?
 What are the revenue and cost drivers of a credit card?
 What is the difference between transactor, revolver and surfer?
 How would you go about building an acquisition strategy for approving/declining
credit card application?
 How would you build a credit line assignment/optimization strategy for a credit card?
Analytics and Data Science Related
 What are the commonly used analytical techniques for Credit Risk Scorecard
development?
 What is rolling, performance and observation window?
 How do you convert probability values to scores?
 What is Points to Odds?
 How do you measure performance of a credit risk scorecard? What the model
performance statistics?
 Explain model monitoring framework for Credit Risk Models?
 What is PSI and CSI?

1. What is a Predictive Model?


2. How do you deploy a Predictive Model?
3. What is difference between classical logistic regression and machine learning
logistic regression?
4. When your target variable is Binary variable,what different techniques you could use
for model development?
5. When do you use Multiple Regression and Logistic Regression?
6. What considerations and steps makes a model predictive during actual application?
7. In a Binary Predictive Model built using Logistics Regression, how do you check if
one variable has higher predictive power than other variable?
8. In a multiple regression model, what is significance of an intercept? Or how do you
interpret an intercept?
9. What is difference between an intercept and an error in Multiple Regression?
10. How is R Square and Adjusted R Square different?
11. What does R Square indicates?
12. What is difference between R Square and Coefficient Determination?
13. When do you use ANOVA and T Test?
14. What is a Predictive Modelling Development Process? Or What steps are taken to
develop a predictive model?
15. How do you select a model for a scenario? What are the model performance
statistics?What are statistics help you decide in selecting a right predictive model?
16. What are the 3-4 statistics used for checking model performance?
17. Consider a scenario, your model does not have good predictive power. What steps
do you take to improve model predictive power?
18. What is model stability and how is it measured?
19. How do you validate a model? What are the different samples you consider for
model validation?
20. A model is ready, what would be your approach to deploy the model?
21. Scenario: Objective is to predictive salary of individuals and salary is continuous
variable. Now available independent variables - Age, Education Level, Gender, # of
Years of Experience, Industry a person is working and City in which a person is
working. Now, some of these variables are non-numeric variables. How can you use
these non-numeric in the model?
22. Scenario: A credit card manager want to target card holders to promote spend on
the card. Aim is find the customers who have higher chances of taking up an offer
and contact them only for the campaign. Which of the modelling techniques can be
used in this scenario?
23. Scenario: Athletes of difference age participate in an Olympic event. We want to test
whether age of athletes winning "Gold","Silver", "Bronze" and "No Medal" are
different.. Which of the statistical tests can be used?
24. Scenario: You are building a binary predictive model, you want to find association
between target variable which is a binary variable and Marital Status which is
Nominal variable. Which of the statistical analysis or statistics can help in finding the
association between a binary and nominal variables?
25. Scenario: A retailer captures details on customers shopping across its stores.
Wanted to understand shopping behavior and segment the customers. Which of the
statistical methods or techniques could be helpful?
1. Model Validation

2. Mumbai, MH

3. FT Permanent

Roles and responsibilities :

- Work with a Leading Bank's Risk Management team on specific projects/requirements pertaining to
Model Risk Management

- Carry out Model evaluations as per the requirements outlined in the Consumer Model Risk Management
Policy for CCAR, PPNR and consumer risk models

- Support MRM team leads in model evaluation of Very High/High/Medium High Risk Level models and
related transactions

- Utilizing SAS, Interthinx (ITX) and other Microsoft Office products, identify key model soundness,
development, and implementation issues, and recommend solutions to mitigate them.

- Perform Independent Validation including but not limited to model eligibility assessment through
updating the model validation reporting templates toward model risk management reviews

- Prepare documentation as per business requirements. Review loss forecasting, stress testing and
macro-economic forecasting models; OR writing model validation reports based on his / her judgments of
the evaluations results.

- Take complete ownership of the project and co-ordinate with customer and all other stakeholders

Required :

- Experience in Credit Risk Scoring / CCAR / PPNR modeling or Validation

- Must have hands on expertise in developing/monitoring/reviewing loss forecasting, stress testing and
macro-economic forecasting models and therefore can evaluate the models developed by regional
consumer teams.

- Strong analytical skills in conducting sophisticated statistical analysis using bureau/vendor data,
customer performance data and marketing data to solve business problems.

- Excellent written and oral communication skills are required, ability to present work in a formal and
understandable format, bringing groups of people to consensus

- Ability to recognizing information and patterns in data that are not obvious, and focusing analytical
efforts in pursuit of explanations, isolations of cause and effect.
- Good programming skills in advanced SAS and SQL in mainframe, UNIX and PC environments

- Strong communication and presentation skills targeting a variety of audiences

- Flexibility in approach and thought process

- Attitude to learn and comprehend the periodical changes in the regulatory requirement as per FED

S-ar putea să vă placă și