You’ve spent all that time collating and storing data from multiple data sources... let’s leverage that data to unlock more analytic insights! We are introducing comparison maps within SpatialKey to empower you to visualize and identify inherent patterns between two points in time, two metrics, or two data sets.
- Accumulation/Portfolio Growth. Given two snapshots of your portfolio, easily visualize and evaluate change over time thematically by region (e.g. state, postcode) or via a heat map depicting concentrations, as measured by any metric in your data (e.g. TIV, AAL). Filter to see trends for a specific line of business, underwriter, business organization, or any other dimension you choose.
- Premium adequacy. Given a portfolio containing AAL and premium information, calculate loss ratios by geography. This highlights areas where your cat premium may not sufficiently cover estimated annualized losses. Visualizing this information will help you identify patterns or concentrations of risks where you assume a disproportionate amount of risk relative to the premium collected. Filter to see how this differs for any dimension of your data, like occupancy, construction, or distance to coast, to identify opportunities to refine pricing and risk selection at the underwriting level.
- Model comparison. Model your data with two different models (e.g. between model vendors or different model versions) and see areas where the models differ most. Again, filter your data to explore more complex relationships - maybe one model predicts higher AAL in the Florida panhandle for wood construction. This will help you formulate your view of risk and built intuition around the methodological differences and views of risk presented between modelers or model versions.
- Market share. Given your portfolio and industry/market data (or even basic demographics, like household income or population), explore your market share by geographic area. Use this insight to identify growth opportunities and to understand where you have a disproportionate concentration of exposure in high risk areas relative to industry.
While all of these use cases are supported, let’s walk through an example that evaluates changes in exposure concentrations from one year to the next using a sample multi-line insurance portfolio. You could also review other metrics, like premium, location count, AAL, etc., or even explore relationships between metrics using ratios like loss ratio, loss cost, market share, etc. And just like all SpatialKey analytics, you can add pods with aggregations of other data attributes and dynamically filter content within other pods and the map, making comparison an even more powerful addition to your analytics toolkit.
Comparison map layers can be presented as heat maps or thematic maps. We’ll start by using a thematic layer where red denotes counties where TIV has increased and yellow where it has decreased. Once we add that layer to the map, we see a thematic map of postcodes in Florida that highlights increased concentrations of exposure. You can hover over any postcode to view a tooltip reporting the magnitude of change and easily add labels for top postcodes that drive portfolio change (shown below).
We can see that exposure changes are more concentrated in coastal regions. The overall picture looks good; however, there are several postcodes that would warrant further investigation given the magnitude of change. You can use this information to further evaluate risk composition in these postcodes using other analytic pods that summarize exposure by business and vulnerability characteristics and identify opportunities to refine your underwriting guidelines.
Additional insight can be gained by leveraging a heat map to tell a more localized story surrounding changes in exposure. Thematic maps are powerful visualization tools to identify concentrations; however, the detailed analysis provided by heat maps allows you to see concentrations without being constrained by geographic boundaries. Here you can see that the barrier islands off the coast of Florida and inland postcodes do have significant localized changes in exposure that gets muted when viewing these as postcode aggregates.