Data Insurance

Sanjib Das

312-***-**** • ********@*****.*** • https://www.linkedin.com/in/sanjib03

EDUCATION

Master of Science in Business Analytics, December 2016

University of Illinois at Chicago (UIC) Chicago, IL

Bachelor of Technology in Electronics and Communication Engineering June 2011

National Institute of Technology Silchar, India

PROFESSIONAL EXPERIENCE

Analytics Consultant, CNA Insurance Chicago, IL, June 2016 - present

Worked extensively on R to create complex predictors and scoring algorithms for insurance comp models.

Performed rigorous input profiling for program optimization. Evaluated model performance using various metrics.

Compare the statistical distributions of predicted loss ratios between multiple subsets of historical data.

Business Intelligence Analyst, Tata Consultancy Services Ltd Pune, India, Jan 2012 - June 2015

Performed data analysis by creating complex multi-dimensional Data Cubes using SQL Server Analysis Service.

Optimized and tuned the performance of data extraction and transformation solution based on Business Intelligence tool (SSIS) for a production environment that eventually decreased the processing time by 30% in general.

Reduced end-users’ effort by 40% by designing a functionality with SQL to capture detailed transit path for each inventory.

Achieved 25% reduction in report loading time by changing the information retrieval from pre-processed to on-demand.

PROJECTS

Classification of San Francisco Crimes - Kaggle Competition Aug 2016

Skills: Python (Pandas, numpy, sklearn), R, Logistic Regression, Naïve Bayes, Random Forests, AdaBoosting, XgBoost

Explored and Visualized patterns and distributions of crime categories with respect to time and geographic locations.

Trained different Machine Learning models with the labeled data and classified 39 categories of crime on the test data.

Achieved top 15% in the Kaggle Leaderboard using best models. (GitHub repository)

App Category Classification Mar 2016

Skills: Python (numpy, sklearn), NLTK, TF-IDF, matplotlib, json

Web scraped textual descriptions of mobile application to get the data corpus. Parsed json object and cleaned the data.

Built customized dictionaries. Classified test app into categories by calculating the cosine similarity. (GibHub repository)

Capstone – MS in Business Analytics

Prediction of NICU Admission – Health Care Services Corporation April 2016

Skills: R (Data table, Caret, ggplot2, rpart), Tableau, Logistic Regression, Random Forests, Chi-Square and P-Test

Performed cleaning of historical claims data, exploratory analysis, missing value imputation and feature selections

Built model to predict the chance a new-born’s admittance into ICU. Achieved 79.5% accuracy using the best model.

Other Academic Projects

1.Built models to predict potential donors to maximize the profit for a direct marketing campaign.

Methods: Random Forest, SVM, Logistic Regression, kNN and Principal Component Analysis. Tools: Rapid Miner

2.Used Clustering techniques to segment the consumers based on their brand loyalty and purchase behavior.

Methods: k-medoids, kernel k-means, agglomerative clustering, and DBSCAN algorithms. Tools: Rapid Miner

3.Developed a collaborative filtering based recommender system that predicts the rating of an item.

Techniques: Matrix Factorization and User-Item kNN. Tools: Rapid Miner

TECHNICAL SKILLS

Analytics: GLM, Lasso, Ridge, Neural Networks, Support Vector Machine, Gradient Boosting, Random Forests, ANOVA, Bootstrapping, Clustering, Matrix factorization, Regularization, Principal Component Analysis, logit, Text mining, Data munging, Feature engineering

Programming: R, Python, SAS, SQL, NoSQL, C#

Tools: Tableau, RapidMiner, Gephi

BI Tools: SSIS, SSAS, SSRS

CERTIFICATIONS: Machine Learning - Regression by University of Washington, R Programming by John Hopkins University

Contact this candidate