All
API
Docs
Demos
Blogs
Webinars

Insurance Fraud Analysis

Explore insurance data for fraudulent activity.

Choose a view below to investigate the data.

Select view

This view shows the raw data model. Although we already see some clustering in this view, by remodelling the data we can reveal further insights.

Select another view from the drop-down menu to see what we can do with KeyLines.

In this view links between individuals are dynamically created based on whether they have an item in common. Individuals are then sized based on betweenness which is a measure of how well connected a node is.

Typical claims are small isolated clusters of people. We can see some suspicious individuals who link multiple claims together.

While it's normal for a doctor to be involved in multiple claims, it's unusual for other individuals - such as a witness, and may be worth further investigation.

This view displays the type of damage caused, with connections to the garages that fixed them. The damages are grouped by car part and links represent nodes with a shared claim.

By grouping the repairs based on the damaged vehicle part, we can see that one garage is repairing a disproportionate number of OS rear doors. This could indicate claim inflation.

Click on a garage to explore the types of damage it has repaired.

Click anywhere on the chart background to undo the selection.

This view places policy holders' and garages' addresses on a map. Links are inferred by identifying paths between garages and policy holders via damages.

By using geocoded data we can see patterns and other information that we couldn't before. Here we see that some people are travelling long distances to get to a particular garage, which could indicate fraud.

It can often be difficult to see the important information in your data, even when everything presented is related.
This demo shows just a few techniques for visualising insurance fraud data to get a better understanding of what insights can be found.

The data is a fictional representation of typical insurance claim data.

Each view combines multiple filters and data manipulations, taking advantage of KeyLines' extensive functionality to create a visualisation that shows different anomalies in the data to help detect fraud. Select a different view from the drop-down menu to see what insights remodelling can reveal.

To visualise data on a map, we integrate KeyLines with ESRI Leaflet and the ArcGIS mapping platform.

Key functions used: