Spatial data analysis is more than just visualisation. Long and weighty books are written about spatial statistical methods, and using them is not for the faint-hearted. In (very) brief, with the right skills at your disposal, the key things they offer your organisation are:
The ability to combine data and analyse relationships between factors
As an example, integrating Census-based demographic data, constituency-based voting behaviour, topographic flood risk data and address-based property value data might reveal information of value. By linking data values at address level, the differences in data geography between the component datasets no longer make the datasets impossible to compare.
The ability to understand local variations
Data for geographic zones, such as local authority areas, electoral wards or TV regions, may offer aggregated ‘average’ values for the whole zone. These have some use, but plainly every business, household or individual within that zone does not represent that average.
Spatially analysing data at finer scales such as Census output areas, postcodes or individual addresses reveals hotspots and common characteristics at a level useful to understand local influences within zones, and how they spill over these theoretical boundaries.
The ability to test the strength of relationships
Robust statistical methods often call for a test of significance – finding out mathematically whether an observed phenomenon is likely to be a coincidence, or if a causal relationship can safely be attributed. Adding a spatial element to a significance test allows us to consider how close in space an observed data point is to the factor which might be influencing, and assign significance accordingly. It’s a ‘real-world’ addition to the statistical methods available.
It’s tempting to treat the analysis of your data as a technical issue, or a statistical one. You need both these disciplines, for sure, but they are the servants of your commercial or organisational need, not the other way around.
In too many enterprises, decision-makers feel blinded by science or ill-equipped to discuss the opportunities and limitations the available data presents. Their technical colleagues may have sound methods, but no opportunity to understand how their work directly serves the objectives of the enterprise.
The result is often IT-led data management and black-box statistics few people understand. The results can be mis-interpreted or fail to meet the original objective, and your investment goes to waste.