Automatic Customer Segmentation
Machine learning as a service for reducing time spent on customer discovery and marketing
Agenda
1
2
3
4
5
Background
Machine Learning Pipeline
Problem Statement
Solution Architecture
Summary
Background
Problem Statement
Automate
Customer Pattern
Analysis
Build and Integrate
with Existing
Systems
Automate Text
Generation
Scale to
Enterprise
Level
AI Vision
1
2
3
4
Machine Learning Pipeline
Data Ingestion Data Exploration
Dimensionality
Reduction
Clustering
Performance
Measurement
Downstream AI
PCA
Kernel PCA
t-SNE
Autoencoder
SVD
K-means
Clustering
Hierarchical
Clustering
Gaussian
Mixture
Model
Silhouette
Coefficient
R
2
Calinski-
Harabasz
AIC
BIC
NLG using
LSTM
Solution Architecture
Summary
Manual content building Semi-automatic content development Approval-only mechanism
80% decrease in manual effort across 4 major marketing campaigns
At least 4 editors + reviewers per campaign
1 editor + 1 reviewer per campaign
1 reviewer per campaign
THANK YOU
