Is the cost of pattern analysis worth the benefits?
What is pattern analysis? Well, its essentially taking the data you collect as an organisation and determining if there are patterns or trends within that data that can help you run your business better, faster, more accurately or more cost efficiently – or simply provide you with a way to make better decisions.
I’m sure you’ve read a lot about big data and the fact that we are creating more and more of it and finding new ways to harness it for the good of our customers. But the pattern analysis I’m concerned with are the stories and nuances that are thrown up by the data we collect over a period of time – whatever that period may be – five, ten, 15 or 20 years. Are the benefits worth the investment over time to capture the data and store it?
Probably the easiest way to demonstrate my point is to look at some examples of pattern analysis as they apply to housing. Geo mapping offers housing associations an interesting dimension to their data. Their estates often span hundreds of square miles, and the ability to make a map from a list of multiple locations, use addresses, postcodes and co-ordinates over time delivers benefits in an interactive map. If there are pockets of anti-social behaviour within the estate over a number of years for example, it’s easier to anticipate and manage it accordingly. Equally, if there are areas where repairs to properties are more prevalent, it may indicate that those properties are affected by flood plains or land patterns for example. In this case housing association may choose to avoid acquiring assets in that area and avoid the associated costs going forward thanks to the knowledge they have acquired from their data over time. These data patterns have the potential to save money – either by flagging up a potential issue that can be avoided or by allowing better management of it with the benefit of foresight.
An example of such foresight might be patterns in tenant arrears. Data may show that residents fall into arrears at the same time each calendar year. There will be reasons for this: a child’s birthday, Christmas or even the winter fuel allowance for example. The ability to predict this means a housing association can deal with the situation in a much more human way. Where they may have chased their tenant hard previously – a text message to alert them may simply be sufficient in the knowledge that it’s a temporary state, or to address the issue through customer contact before it occurs.
We have to get much better at making our data work for us and for our customers – whilst its costing us to store it, we should be making the most of mining it to deliver us the knowledge to work smarter.
This brings me neatly on to the issue of machine learning and Artificial Intelligence. The premise of AI is that computers will eventually do the things that require intelligence when done by humans. Whilst machines are excellent at finding patterns in data that can tell the story – and can undoubtedly already do it faster than you or I can – they are still very poor at being able to tell those stories. So whilst AI is becoming a reality, there is no way to remove the human element (for the time being) from our pattern analysis. With time that may change, but intuition and persuasion are still very human qualities and when we’re dealing with customers, those matter enormously.