Customer Clustering

This Shiny app is designed to simulate how a hypothetical company might analyze its customers based on their similarities.

You can see the live app On my ShinyApp.io account. Refer to the Shiny documentation for help with running the application.

Background

This app is designed to show how clustering techniques can be used by an organization to examine similarity across their customers. The primary motivation is to group customers by behavior to streamline initial pricing processes, but it also includes tools to examine purchasing patterns of customers based on the types of products they buy (Product Class) and the organization's market position in terms of competition.

Context

In this scenario, the organization has several dozen customers, each buying some assortment of their hundreds of products. The products each belong to one of 12 classes Product pricing is primarily driven by customer size in terms of a discount off of the list price and rebates based on their throughput. Since the top 7 customers make up over 75% of the organization's profits, they are interested in grouping their smaller customers to streamline pricing processes.

Also of interest is a tool to allow the organization to compare customers in terms of the Classes and Exclusivity of the products they buy, potentially identifying areas of risk and opportunity for each of the customers.

Components/Tools

All Customers

The All Customers section provides a high-level overview of all of the organizations customers, including:

A heatmap of all customers, allowing for a quick identification of customer grouping using Hierarchical Clustering on the main four customer attributes
A 3D Scatterplot showing the distribution of the customers according to how they cluster using K-Medians clustering (with the Top 7 grouped separately), along with a summary table of the groups
Sections for comparing customers to each other and the organizations overall distribution of Product Classes and Exclusivity. These sections include a visuals of the organizations aggregate distribution, heatmaps of the customers distributions, and a tool for comparing each customer to the aggregate distribution.

Small Customers

Focusing on the smaller customers, this section provides:

A similar heatmap to the one in All Customers
A 3D Scatterplot of just the smaller customers, including bloxplots detailing the differences in the clusters, group summaries, and details of the top customers and products within each group.

Product Views

A 3D Scatterplot of the organizations portfolio, grouped by Therapy Area. A wide range of variables are available for the axes.

raffled / customer-clustering Goto Github PK