Aman Arora Data Scientist

Data Scientist

Anheuser-Busch InBev

Apr 2021 - Current (5 years 3 months)

Barley Price Forecasting Model • Developing a model to forecast weekly Barley Prices using exogenous features like weather, fx rates, other commodity prices etc. • This will help the Global Procurement team in taking a decision on when to book the Barley volumes. E-Auctions Model Development & Deployment • Contributed to E-Auctions optimization model development with deep dive analysis on model features. • Deployed the model on AzureML cloud services as a web service. Managed a vendor for front end development changes for this deployed service.

Assistant Manager (Analytics)

Titan company ltd

Dec 2018 - Apr 2021 (2 years 5 months)

Dormancy Winback Propensity Model • Developed a dormancy winback propensity model to for the watches business using an ensemble approach of techniques like Logistic Regression, Random Forest, XGBoost. • Model performance - The conversions in the first decile were around 50% as against 10% generally. Business Impact: The business was able to generate a winback revenue of ~2.3 cr. over a period of 2 months. Customer Segmentation • Developed Segments of watch customers using K means clustering algorithm with a profiling layer on top of it. • Segments like Loyalists, Bargain Hunters, Festival Buyers etc. were identified. Silhouette Score was 0.5. Business Impact: These segments were used for multiple campaigns to target loyal customers,

L

Manager (Analytics)

LEE FINANCIAL SERVICES

Jan 2018 - Nov 2018 (11 months)

Collection Propensity Scorecard • Developed a propensity model to identify customers who are more likely to default in the next month for the Two- wheeler lending business. The conversions in the first decile were around 13% as against 1.32% generally. AUC value-0.72. Business Impact: The model enabled the collection teams to prioritize their efforts towards customers with high propensity to default. Employee Attrition Model • Developed an employee attrition propensity model using Logistic Regression. Model performance- The conversions in the first decile were around 73% as against 22% generally. Model lift- 4.95, AUC value-0.70. Business Impact: The model gave insights to the HR team on the factors responsible for employee attrition

H

Management Trainee

Hansa Cequity Pvt. Ltd

Apr 2016 - Jul 2017 (1 year 4 months)

Cross Sell Propensity Model • Developed a Cross-sell propensity model to identify customers who were likely to buy a particular product from the existing base. The conversions in the first decile were around 2.5% as against .76% generally. Model lift- 3.5, AUC value- 0.72 Up Sell Propensity Model • Developed an Up-sell propensity model to identify customers who were likely to upgrade to the premium product from the existing base. The conversions in the first decile were around 2% as against .51% generally. Model lift- 2.9, AUC value-0.74.

Aman Arora

About

Detail

Timeline

Résumé

Aman Arora new_releases

About

Detail

Aman Arora