Business Analyst

Kawal Deep Singh

Team Lead, Axtria India Pvt. ltd.

Email id : ************@*****.***

Contact # : +91-989*******

Summary of Professional Experience

Rich experience in statistical analysis, predictive modeling, profiling, segmentation and cluster analysis, fraud analysis

Experience in modelling techniques like Logistic & linear regression, ARIMA time series forecasting and decision trees. Proficient in data handling and manipulation of large data sets using SAS, SQL Server and Hadoop, Teradata etc.

Base SAS certified SAS Programmer and Extensive experience of SAS programming for automating various task in all the projects worked on.

Worked in insurance, Banking, Utilities and Healthcare domains.

Total experience of 5+ years

Handled a team of 6+ team members

Have been promoted twice within the span of 3 yrs. in last organization.

Received award as Employee of the quarter for year 2012 Q3 and 2015 Q2 in EXL Services (formerly Inductis) and always been top performer in the team. Received appreciation from client top leaders for doing good statistical analysis and predictive models.

TECHNICAL SKILLS:

Programming : SAS (Base SAS certified), R, VBA

Analytics Tools : Enterprise-miner and CART Analytics tool

Methods Implemented : Logistic, Linear Regression and Decision Tree etc.

Operating Systems : Windows, UNIX

Documentation : MS Office (Word, Excel, Power point, Access,

MS Project)

Career Summary

Nov’15 - Current Axtria India Pvt. Ltd., Gurgaon, India

Jun’12 - Oct’15 EXL Decision Analytics, Gurgaon, India

Mar’10 - May’12 Tata Consultancy Services Ltd.

Education: B.tech in Electronics & Communication, 2009, IEC-CET (UPTU) India

Career History:

Nov’15-Current

Company: Axtria India Pvt. Ltd.

Industry: BFSI

Credit cards risk analytics (retail and commercial)

Worked with retail/commercial credit cards product teams to devise strategies from projects related to reducing net losses and curtail risks 30+/90+ delinquencies (by early prediction and monitoring over time)

Net credit loss forecasting using flow rates for different commercial card portfolios using ARIMA technique

Default score prediction for credit cards customers by building a logistic regression based model followed by validation (out of time and bootstrap)

Delinquency and vintage curve analysis for projecting portfolio performance in reviews with the product teams to come with further actionable

Wallet share analysis for clients by estimating their wallet size through RTGS transactions done with other banks and comparing it with NRV (net realizable value)

Spends and utilization analysis to divide the customers in segments and make marketing/campaign strategies accordingly

Profitability analysis for commercial banking clients basis cost on various financial transactions and float revenue

Credit cards collections analytics of private label credit cards (PLCC)

Worked in the collection analytics strategy team for the Private Label Credit Card division of a US based major finance company

Collections strategy: Part of the collections strategy team, responsible for designing champion challenger strategies for optimal allocation of collections efforts

Optimizing collection channel attempts: Building database of the entire attempts/contacts activity on the delinquent cardholders through different channels and all the payments made. The database was then looked from different angles so as to take a stock of what attempts are causing what payments and will thus help optimize the collection efforts

Mar’10-Jun’12

Company: EXL Services

Industry: Insurance

Project# “Customer Retention Model”

Language/Tools: SAS, E-Miner, Microsoft Excel

Client: Leading Life Insurance Major in India

Team Size: 2Progammers and 1Team Lead

Role: Senior Business Analyst developed the model from scratch using logistic regression method and presented to the client. Worked with IT on implementation of this project on websphere .

Project Details: Objective for this model was to develop a retention model which can assist the company by identifying key drivers for increasing retention and focusing the communication to the "right" customers.

Key business drivers affecting the customer retention were studied and the characteristics of all probable variables at product / channel, Agent, and customer level were drawn.

Data profile of Customers was segmented according to key attributes based on historic behavioral pattern & demographics.

The model was developed in a step-wise fashion, introducing the significant variables one-by-one into the model in the order of their significance.

Different parallel models were developed using decision tree rules, logistic regression & as well as neural network in SAS/STAT & E-Miner analytical tools.

Performance of these models was tested on an out-of-time data set, and then finally logistic regression model was selected as the one capturing maximum responses with least population.

K-fold technique was employed for creating out-of time validation set due to limited data availability in the data environment of one of the channels.

Project# “Early Claim Prediction Model”

Language/Tools: SAS, E-Miner, Microsoft Excel

Client: Leading Life Insurance Major in India

Team Size: 3Programmers + 1Team Lead

Role: Business Analyst worked in data preparation, clustering, bivariate analysis. Developed the model on E-Miner and did out of time, out of sample validation .Worked with IT on implementation of this project on websphere.

Project Details: The pattern of early policy claims to be empirically investigated with a view to formulating a model which will capture the early-claims phenomenon.

Primary goal was to predict before issuance whether a policy is risky from early claim (claim with in a year of policy issuance) perspective or not. Based on this prior insurance activity of both the agent and the customer a lot of variables (both on the customer and agent profiles) were created considering all the transactional, demographic and insurance behavior.

Bivariate analysis was done over all the categorical and continuous variables

Based on these exploratory data analyses, some more ancillary relevant variables were created and subsequently a model-able partition of the population was identified through conventional segmentation procedures.

Tried many algorithms like Decision Tree, Rule Induction and Logistic Regression for model building. However we opted for Decision Tree as it gave consistent and accurate results.

Project# “Persistency Projection and Planning”

Language/Tools: SAS,E miner tool, Microsoft Excel

Client: Leading Life Insurance Major in India

Team Size: 2Progammers and 1Team Lead

Role: Business Analyst.

Project Details: Objective of this Task was to project business persistency for next year by studying historical trends of both persistency and share mix of various divisions.

Persistency for each channel is built bottoms up using ‘tree approach’, studying the effect of each key driver.

Key drivers’ up-to levels 6 were studied to find underlying patterns in persistency movements.

Persistency across different points in time were studied to establish trends

Further, deep-dive was performed to find out the change in the persistency trends

In order to make assumptions as close to reality as possible, last 6-8 months collections trends were studied and thus lifts in persistency were assumed

.

Mar’10-Jun’12

Company: Tata Consultancy Services

Industry: Clinical Statistical Analyst

Worked with Tata Consultancy services for a Leading US based pharmaceutical client and responsibilities revolves around decision making through use of quantitative data analysis and then creating Tables, figure & Listing (TFL) specification from Clinical Extract datasets and Analysis Datasets

Project#4 “ADM attrition and consultant performance Analysis”

Language/Tools: SAS Drug Development (SDD), PC- SAS, Excel

Role: Clinical Statistical Analyst

Project Details: The client is a leading pharmaceutical firm in the world. The project revolves around the creation of Tables, figure & Listing (TFL) specification and creation of Tables, figure & Listing from Clinical Extract datasets and Analysis Datasets.

Specification Creation:

Creating requirements for Statistical Analysis using various Study documents

Specifying the Statistical Methodology to be used, transformation, derivation of new variables.

Standardization of Mock-ups library for the Statistical analysis reports.

Creation of Metadata which can be used for Programming.

Statistical Programming:

Analyze the requirements and develop code accordingly using SAS programming and generating the report in rtf format using Data NULL and Proc Report.

Creating tables, figures and listings various Statistical Techniques for Data Modeling.

oExtensive use of Advance SAS techniques Like SAS STATS, SAS GRAPH, SAS MACROs, SAS SQLs,

oAdvance Optimization techniques.

Creating various like Scatter Plot, Line Graph, Box Plot, and Bar Graph.

Testing and Validation of programs.

Contact this candidate