Barkha Bansal

Utilized unstructured Price and Cost Time series data to prepare metrics for Price Optimization. Responsible for automating and improving elasticity modelling pipeline using UCM. Responsible for automating data engineering pipeline as initial step for Price Optimization, including various data formats, data cleaning, data modification, working with large data.

MMM across Traditional media, Digital Media, Trade and Consumer Promo. Time varying coefficient using DLM in R, incorporated in Python pipeline in Azure. Nested Hierarchical Multiplicative Mixed Model for nested hierarchy (LME4 - R and Statsmodels - Python). Baseline estimation techniques/ Time series decomposition: Holt Winters, UCM, Spline Interpolated de-trended relatives as seasonal indices. Parallelisation of UCM (Asyncio). Bayesian Regression (PYMC- Python) to overcome limitations of OLS and include prior domain knowledge/ information. Web-based simulation and optimisation tool/ app for Australia market on Azure. Present complex findings to non-technical stakeholders using automated decks (PPTX), web-based App, automated excel reports (VBA). Owning POC: Global solution scaling with a uniform product hierarchy: Homogenized Item Attribute Database using text analytics (Regex, similarity, matching, rule based criteria) (PySpark.ML.Feature). Successful POC including 2 categories and 2 BUs resulted in project extension of up to 12 US BUs, 4 CA BUs and 10 EU BUs. Technical training resources including Pyspark, R and SQL, text analytics, automated excel reports (VBA macros). Parallelised baseline estimation using UCM as first step to Price Optimization, followed by pipeline creation and automation. Automating price cost extraction from various unstructured sources across 15 BUs for Price optimization.

Delivered a quality learning experience for various courses lie BSc Mathematic Hons, BCom Hons, MSc Applied Mathematics, with the aim of holistic growth which was met with consistent excellent faculty feedback.