Sign in

Data Sales

Hartford, CT
March 18, 2019

Contact this candidate


Poojith Goud Routhu

860-***-**** 915 Main St Hartford CT 06103 LinkedIn GitHub EDUCATION University of Connecticut School of Business Hartford, Connecticut M.S in Business Analytics and Project Management, CGPA:3.7/4.0 Aug 2018 – Expected Dec 2019 Birla Institute of Technology and Science, Pilani-BITS Hyderabad, India M.S in Mathematics, CGPA:3.7/4.0 June 2012 - May 2016 DATA SCIENCE COMPETENCIES

Analytics skills: Clustering, Machine Learning (ML), Time Series Models, Statistical Modeling, Predictive Modeling A/B testing, Deep Learning, SVM, Boosted Trees, Regression, Naïve Bayes, NLP, Data Mining BI Tools: Google analytics, Adobe analytics, JMP, Azure ML studio, Octave, MS Excel, ETL, MS Power Point, MS Project Data Visualization: Tableau, Power BI, DOMO

Libraries: dplyr, CARET, ggplot, scikit-learn, NLTK, RShiny, pandas, numpy, TensorFlow, matplotlib, Keras Programming: C, C++, Java, Teradata, MATLAB, PYTHON, R, SAS, SQL Math skill set: MathWorks, multi-variate analysis, Hypothesis testing, statistical analysis, ANOVA, MANOVA, Optimization WORK EXPERIENCE

LATENTVIEW ANALYTICS, Data Analyst, Chennai, India June 2017 - July 2018 Defective Transaction Fraud model for eBay client

• Data Engineering Set up an ETL process to obtain, clean and organize data using SQL, R and Excel. These extract transactional & text conversation data between eBay buyers and sellers segregated at user level data

• Predictive Modeling NLP Text Mining Developed machine learning models like Naive Bayes, SVM and CNN in R and python on a sample of 600,000 conversations and transaction data with an accuracy of 87% and higher recall value

• Data Visualization Predicted defective transactions increased the global collections gross profit margin by 13.8% by reducing drop shipment costs. Generated reports and provided business insights to clients for initiating action over quantitative transactions at a geographic level through various channels by creating interactive tableau dashboards Reporting and ADHOC for eBay client

• Data Engineering Set up an ETL process to obtain data by building report codes in SAS(Macros). This extracts different transactional data (i.e. Product Invoice, price, roll rates, site) to Teradata tables aggregated at the user_id level

• Automation Used SAS and SQL scripts to “automate key regular reports” thereby reducing the turnaround time.

• Quality assurance & statistical data analysis for specific business requests. Created tableau dashboards for insights and recommendations on daily and monthly transaction data using roll over rate metrics.

• Data Modeling Business intelligence Built fraud suppression model for the eBay global collections team using logistic regression and boosted trees in R & Python by considering variables of business impact with frequent client discussions. Calibrated parameters of the model for business impact. Implementation of this model increased profitability by 10.75% in the area of unpaid items sector, reducing transaction data for unpaid items over a period of 4 months JDA SOFTWARE SOLUTIONS Business Analyst Supply chain management India June 2016-June2017 Forecasting sales and Inventory Optimization for Walmart client

• Modeling Built a tool to forecast the winter promotion sales for Walmart using the Bass diffusion model and regression.

• Increased sales forecast accuracy by 12.5% using time series models (ARIMA and exponential smoothing)

• Analyzed trends in sales for various promotions using an Optimized algorithm in ERP System

• Developed models of Decision Tree Classifier, Random Forest & generalized regression models on 30,000 users and provided recommendations on the business strategy for optimizing supply chain costs for $12Million budget ACADEMIC PROJECTS AND COMPETITIONS

Customer Analytics Google Analytics Customer Revenue Prediction Predictive Modeling R Python Tableau Analyzed Google Merchandise Store customer dataset and provided recommendations for better use of marketing budgets by predicting the revenue generated per each customer. Techniques used: GBM, Neural networks & logistic regression. Achieved AUC of 94% Fraud Analytics 2018 Travelers Statistical Modeling Competition Predictive Modeling Python JMP Identified first-party physical damage fraudulence explaining reasons of fraudulent claims based on historical transaction data Achieved AUC of 75% Among Top 100 on the Leaderboard for the best models Customer Analytics Telecom Churn Prediction Predictive Modelling R JMP Tableau Predicted whether a customer is going to churn out or not with an AUC of 84% and also determined factors causing churn for a telecom provider company

Contact this candidate