Shalini E.M

Statistics

Specialization : Actuarial Statistics and Time Series Analysis

Physics, Mathematics and Statistics

Physics, Chemistry, Mathematics, Biology

Business Analytics

• Customer Complaint Analysis(Jakarta, Indonesia): Identify the top complaints from external feedback in near real time (twitter, boards, reviews) with IBM Watson Social Media Analytics.
• Identity Matching(Jakarta, Indonesia): Identifying customers who transfer out money to another bank with dissimilar names using probabilistic matching algorithms and develop real time scoring with CADS.
• Predictive Maintenance (Singapore): Analytics-based early warning and root cause analysis system for temperature forecasting of servers during stress testing and optimizing test cell allocation to reduce heating of servers for manufacturing client.
• Geological Modeling of Mineral deposits (Manila, Philippines): Predicting the presence of iron and nickel deposits for a mining client and sharing the business practices.
• Telecom Customer Affinity Analytics (Bangkok, Thailand): Predicting the post paid customer’s affinity for buying certain products and services for a major global telecom provider.
• Customized SPSS Modeler and SPSS Statistics workshop(Singapore, Philippines, Thailand, Vietnam): Conducted Analytics workshop for clients and business partners in the areas of churn analytics, predicting defaulters in a banking scenario and conducted Standard SPSS trainings for banking, telecommunication, pharmaceutical, mining clients/business partners.

Policy and Compliance Management Solution for Asset Management Company (Mumbai, India)

Proof of concept for automating the policy management life cycle to achieve compliance, mitigate risks and ensure adherence to corporate policies and procedures.

Operational Risk and Shariah Risk Management (Kuala Lumpur, Malaysia)

Lead the first Open pages implementation in ASEAN for a leading bank by collaborating with Chief Risk Officer, Head of Operational Risk Management, Head of Policy Management in implementing and solution designing for Incident management and action tracking of losses occurring due to operational failures, Risk and Control Self-Assessment, Key Risk Indicators and training the entire Operational Risk Management team in using the solution and the product.

Operational Risk and Internal Audit Management (Bangkok, Thailand)

Lead the solution design for a bank for the office of Chief Risk Officer and Head of Internal Audit by understanding the business problem, designing the solution for incident management and action tracking, risk assessment and control effectiveness, audit scoring , audit initiation to completion.

Presales Support:

Worked with presales and practice managers to estimate the scope of work, resource estimates, support for response to RFP's to win both the clients above in Malaysia and Thailand.

Worked as a statistician, functional consultant, business analyst, subject matter expert and single point of contact for advanced analytics, functional design of Credit and Operational Risk solutions and numerical algorithms.

The core solutions that I managed during my tenure here are:

Retail pooling solution (July 2007- Jan 2008)

Under the Capital adequacy framework of Basel II, banks will for the first time be permitted to group their loans to private individuals and small corporate clients into a ‘Retail portfolio’. As a result, they will be able to calculate the capital requirements for the credit risk of these retail portfolios rather than for the individual accounts. I joined this solution during system testing phase .During this period, we had tested, modeled and released the retail pooling solution.

• Data analysis
• Regression Models
• Creation of pools using Hierarchical, k-means clustering.
• Variable Reduction using Factor Analysis
• GINI
• Entropy
• Impurity Measure
• Report Generation-Pool Stability Report.
• Basel Output Report

Credit Risk Economic Capital Solution (Jan 2008- Jan 2009)

Functional design, statistical algorithms, quality testing of Economic Capital Solution aligning to Basel II regulations. The out of box  models were developed on the probability of default (PD), the loss given default (LGD) and the exposure at default (EAD).The other models I worked are:

• Time to default Model
• Merton Model
• Cash Flow Models
• MCEM model, Transition Matrix, Credit Metrics Structural Model.
• Conditional Default Model
• Black Scholes Model.
• Historical Default Weighted Average of Pool Observed LGD
• Distribution Fitting
• Logistic Regression
• Credit Metrics Structural Modeling
• VaR, cVar, Expected Loss, Monte Carlo Simulation, Unexpected Loss, Economic Capital.
• Data Model understanding and data preparation, loading and modeling
  Functional testing of the models
• Test cases for specified applications
• Test, record and report defects in testing a software application
• Provide technical design for product enhancements

Operational Risk Economic Capital Solution (Jan 2009- September 2009)

The new accord emphasizes the financial institutions to make Operational Risk assessment as one of their integral component of risk management system. ROREC product calculates economic capital for operational risk using the Loss Distribution Approach as well as Scenario Analysis and application of Standardized Approach for Operational Risk. The major areas of the solution that I worked are -

• Severity Modeling for Scenario Data.
• Distribution Fitting for 20 distributions
• Functional testing, Test Strategy Preparation, Test Management
• Monte Carlo Simulation result validation for 20 distributions.
• Validation of the results from NAG libraries.
• Frequency of Loss Modeling.
• Copulas
• Unexpected Loss, Expected Loss, Value at Risk, cVaR calculation for Regulatory and Economic Capital.
• Operation Risk Reporting
• Insurance Models–Proportional Retention Model, Aggregate Deductible Model
• Data Model Understanding
• Data Model understanding and data preparation, loading and modeling
  Functional testing of the models
• Test cases for specified applications

Stress testing and CREC‘s OBIEE Dashboard for  ICAAP (September 2009 – Jan 2010)

Stress Testing framework to enable risk managers to measure losses from extreme, although plausible, scenarios Stress testing exercises are widely used by financial institutions in assessing their exposures to credit and other risks. Stress tests can also help policy makers to gauge the potential implications of differing risks for the stability of the financial system as a whole. And in recent years, there has been a burgeoning interest in such systemic stress testing among central banks and international organizations

The major areas that I worked are

• Building models to gauge Macroeconomic stress tests
• Stress Testing PD , LGD and EAD models
• Future Value Shock
• Monte Carlo simulations and stress tests
• Estimation of Baseline distribution
• Models for simulated frequency distributions of credit loss under baseline and stressed scenarios
• Credit Risk Economic Capital Stress Testing
• Data Model understanding and data preparation, loading and modeling
  Functional testing of the models
• Test cases for specified applications
• Test, record and report defects in testing a software application
• Provide technical design for product enhancements
• Stress for bank’s Mortgage Portfolio
• Interest Rate Shock Model

Advanced Analytics Infrastructure (Jan 2010 – July 2011)

Involved in conceptualizing enhancements to the Reveleus Advanced Analytics Infrastructure that was also part of second version of Operational Risk management. A key advisor in developing in-house graphical functional for Statistical Models using gnuplot, pre sales consulting,training the development and pre sales groups on risk systems, products and processes and the way these processes are achieved at a software level

Solvency II (July 2011 – Nov 2011)

Solvency as referred to by the EC Insurance Directives means the financial resources of an insurance undertaking, i.e. in essence the difference between the assets and the liabilities of the insurer. This kind of safety capital is necessary in order to absorb discrepancies between the anticipated and the actual expenses and profits

• Requirement understanding
• Preparation of Functional Design Documents
• Analyzing the Quantitative impact Study , Reports
• Functional Inputs for Data Model
• Analyzing various Risk Drivers , Reinsurance , Run off triangles ,Group insurance reports

Implementation and Presales Support

The products designed and developed by OFSAA, Bangalore has been in the Gartner's leader's quadrant. Being the only statistician in the team, I  had provided my expertise and advise in winning world wide clients for solutions. Being part of a product life cycle is a satisfying experience. I worked with sales, pre sales globally to win deals in India, North America, ASEAN , Australia , Middle East. Post sales, I was actively involved in guiding the implementation team, product support teams till the go live of the solution

Here I joined as a fresher SAS trainee with hands on experience Credit Risk modeling and Marketing management using SAS BASE, SAS MARO and SAS Enterprise Miner.

Mixed Response Modeling (September 2006 - January 2007)

 A cola manufacturing company pilot surveys products to strategically place advertisement for its various flavor, size of the products before launching three products in to the market .The ads are placed in different newspapers across the state. Now the company wanted to know which publication, edition, offers and day (or as combination) is fetching more responses.

Typically, the marketing unit of a company would like to identify the marketing drivers (such as price, promotions, distributions, amount spent on advertisements, different forms of media (say, TV, print, outdoor, radio, etc.) that influence in enhancing or declining sales or awareness of brand / category. The general objective would be to understand the simultaneous effect of their own sister brands (Halo and Cannibalization effect) / categories and other competing brands / categories that affect their sales or awareness. The responses are in terms of total calls received. The objective is to use as much information as input to these models in order to establish which factors are the most predictive. The objective is to identify the best combination of factors. Modeling the relationship between response variable and explanatory variables, also to get the best combination among explanatory variable which influencing the response variable.

Analytical Models used

• ANOVA with Interaction effects
• Tukey's Test
• Test For Homogenity of Variance
• Exploratory data Analysis
• Cluster Analysis
• Factor Analysis
• Generalized linear Models

Credit Risk and Customer Relationship Modeling (January 2007 – April 2007)

A credit card company wanted to know which card, offer is most popular among the customers.

 This group of Analytics focus on Cross-Selling, Campaign Design, Execution, Campaign Response and Target List Generation. Their purpose is to identify cross-sell and up-sell opportunities, generate lists for action, help focus campaigns on specific segments and products which have highest probability of success, and analyze the effectiveness of the executed campaigns, so as to improve the design of future campaigns. Additionally, some of the Marketing Analytics will now include analysis of Prospect Databases.

This analytical domain permits analyses of the Product Level Holding of all customers by their demographic dimensions to identify new customer segments, as well as hidden trends, which may be profitably exploited by introducing new products and services. This analytical domain also facilitates comparing one demographic feature across another, so that the entire portfolio can be analyzed against various demographic combinations.

This analysis leads to answers to the following strategic questions:

Which Product is most favored by the bank’s customers? What is the profile and source channel of such product holding customers?

Which Products are the least popular with customers?

Statistical Techniques used

Multivariate Analysis

Trend analysis

Behavioral analysis

Risk Profile analysis

Attrition analysis

Cluster, Discriminant, GLM, CHAID

Campaign analysis and Segmentation Analysis

Campaign Response rate

Campaign Approval rate

Campaign Acquisition rate

Adjustment Factor for all campaigns during the average life of the account