LG Electronics Production ML
Customer intelligence platform
Customer lifetime value and next-basket models, delivered as a working product for retention and e-promoter teams.
- customers scored daily
- 35M+
- top-5 accuracy
- 80%
The problem
Customer outreach needed a practical way to identify what someone was likely to buy next, what they might spend, and how likely they were to purchase within the next year.
What I built
Combined a two-stage customer lifetime value model with a next-basket model spanning 50 product categories.
Produced daily predictions for the next purchase category, expected spend, and 12-month purchase probability.
Delivered predictions through a production interface used by retention and e-promoter teams to inform outreach.
The outcome
The platform scores more than 35 million customers daily with 80% top-5 accuracy. It supports programs projected at more than $12 million annually; that figure is a projection, not realized revenue.
A summary of the work described on my public profile.
View on LinkedIn