Wharton Customer Analytics Datathon February 2021
After exploring the characteristics of Essity's Tork data sets, we identified potential areas of improvement in reducing wastage and increasing the efficiency of customers' cleaning teams.
The motivating framework of our analyses is a 2x2 matrix, with Traffic (below-average or above-average) on the vertical axis, and dispenser Status (empty, full) on the horizontal axis. We call this table the Essity Utilization Matrix, which provides an interpretable, actionable scheme for understanding and improving customers' operations. Each cell in the matrix (ranging from "bad service" to "sweet spot") directly maps to the business context. Our goal is to predict daily traffic within a given location to proactively prevent dispensers from being under-stocked on those days.
The key metric in our analyses is dailyTraffic, a feature we engineered using the Tork data sets. This variable served as the response in our regression analyses (multiple linear regression and elastic net) and helped motivate our classification study as well.
Overall, using these frameworks, we found meaningful results as to which covariates best explain and predict daily traffic.
We are excited to share our code and presentation with you!
Log in or sign up for Devpost to join the conversation.