Jivisha Kumar

Data Science and Analytics Team Key Projects: COVID-19 Pandemic Research Due to the widespread outbreak of COVID-19 , there is a significant impact in U.S financial market. Transunion participated in numerous analysis to calibrate the impact and development of predictive models and algorithm to mitigate credit risk and financial strain. Value added in project: •Collaborated with third party survey data & used N.L.P techniques in Python to analyze the of ~5M affected population •Created 40+new COVID related algorithms (Credit Vision Acute Relief Recovery) for efficient risk mitigation •Identified ~41.3M consumers under COVID-19 accommodation and evaluated changes in Payments, Credit Card balances and consumers liquidity. Improved delinquency rate by 142 bps as of June 2020 Propensity Model for Mail Campaign: Increased model’s AUC by ~30% compared to Generic Personal loan propensity model by building segmented data • Used K-Folded Extreme Gradient Machine Learning technique for modeling and Cross Validation for parameter tuning • Created segmented audiences using propensity model to target potential loan buyers Reject Inference for Risk Model: • Estimated likelihood of default for the rejected and not-booked applicants using CGLM. Identified important attributes using Ridge & Lasso regularization and engineered new features for model. Developed model with K.S value of 72% • Leveraged model output to analyze risk for through-door applicants. Reduced delinquency rate from 8% to 4.5%

Data Extraction: Created pipelines to extract data from excel and pdf documents. Built data schema to store data in MS SQL server Root Cause Analysis: Analyzed root causes for HERE map errors & wrote views in MS SQL for data extraction Automated Report: Summarized map performance in dashboard for executives using Tableau saving 30% of operational time

Application for Model Evaluation and Portfolio Segmentation using SHINY R: • Automated process of model evaluation using SHINY R, used Gain Chart & Population Stability Index for stability and robustness check for in time and out of time sample. Thereby reduced team’s effort by 10 hours/month •Designed evaluation reports in R using Knitr and RMarkdown • Upgraded approach of portfolio segmentation using CART Decision Tree ML Algorithm. Thus reduced delinquencies by ~10% Process Automation of retrospective analysis: • Revamped and automated bulky process of extraction and cleaning million or records from data present in different platforms using Informatica & Shell Script. •Overhauled analytical steps from SAS to R, thereby by reconciled time & cost on SAS licensing and processing time by 60% • Created user defined functions in R using Dplyr, Magrittr, Ggplot2, lubridate & Stringr to segment population on basis of Value and Portfolio & statistical analysis (K.S value, correlation value).

- Automated weekly reports for supervisors using Excel-based dashboard and Tableau. - Used platform of Google Web Analytics to monitor web traffic and produce insights about the department's website viewers. - Maintaining database and making necessary modifications. - Assisting the Department of Urology technical Center’s IT Technical Associate with the maintenance of web projects using Wordpress and HTML

• Designed and developed web-based application and auto-reports to monitor critical indicators for a performance of crew members and their in-flight services. • Used Power BI Tool, SQL Management Studio and Azure analytics for report creation and analytics. • Wrote Power Queries, DAX calculations for the data analysis • Worked with business analyst and customers in an agile framework to realize business use cases into user stories. • Saved 40 hours of time spent on data pull by establishing automated data pipelines and creating reports for data governance. • Increased data movement and utilization by 35% without needing any additional resources. • Delivered analytics derived business solutions for various critical business scenarios. • Managed reduction of heavy workload for business users by 60 percent on manual paper evaluation and report creation. • Awarded twice as a Best Performer of Sprint.

• Worked on POC (Proof of Concept) creation for travel, transport and hospitality customers for their various business scenarios. • Developed statistical-based solutions in R using Linear Regression and Random Forest Algorithm for ticketing and booking analysis. • Worked in developing a predictive model for booking and cancellation pattern for the Hospitality Industry in R using Random forest algorithm with h2o package. • Worked on Predictive Sales Model for the logistics customer and report creation using Shiny Application in R. • Developed encryption and decryption framework using Java in Talend DI -ETL Tool, for secure data ingestion from client environment to development. • Worked on graphical data analysis and database management in Neo4j for aviation customers

COURSE WORK -Business Statistics -Data Mining -Statistical Modelling -Machine Learning -Natural Language Processing using Python -Advance Database Management -Applied Consulting -Data Visualization with Tableau -Accounting

Coursework: Electronics & communication , Control System, Embedded System, Robotics, Statistics, Applied Mathematics