We often hear talk of how analytics can help businesses and organisations extract valuable insight from their data, but that’s where the conversation ends.
How does it do this? And what should we do with the resulting information?
Analytics can be applied to any scenario. It can help to minimise wasteful expenditure within a government department, or to improve tax compliance among 30 to 40 year olds, for example.
Whatever the application, data analysis provides useful insights that can be leveraged to not only make strategic business decisions, but also to solve problems, replicate desirable circumstances, or prevent issues from occurring in the first place.
When discussing analytics, we refer to predictive analysis, descriptive analysis and sequencing, which each employ various analytical disciplines, technologies and methods.
One of the fundamental tasks of analytics is to bring predictive power to information. It can help us to understand what happened in the past, what is happening now, in terms of monitoring and controlling information), and what is likely to happen in the future.
Descriptive analytics is about understanding certain phenomena. In any industry, there are those who want to be able to detect two types of patterns – desirable and undesirable.
For example, a hospital chain may detect that Hospital X is performing well in terms of disease eradication, which is an example of a desirable pattern. The hospital group may want to drill down to find out what Hospital X is doing differently compared to Hospitals Y and Z, with a view to replicating this model across the chain. Similarly, it may notice that Hospital Z has a worryingly high infant mortality rate – an undesirable pattern – and, through analysis, will be able to identify what factors contribute to this.
When analysing a sequence of events, one tries to deduce what order of events produced a certain desirable or undesirable outcome with a view to preventing something from happening, or ensuring that desired results are repeated.
With this information, the group can rank hospitals on a good/average/bad spectrum and settle on a benchmark that all hospitals can be measured against. Once it understands the variables around those hospitals that are performing well versus those that are not, it can then implement solutions at under-performing hospitals, and can replicate models of successful hospitals, in the process preventing disease outbreak and lowering infant mortality – before it happens. Ultimately, analytics is about inducing general, non-specific knowledge that can be applied elsewhere.
Crucially, organisations must act on the insights produced by analysis. However, they will not come by this information overnight. Data must be aggregated over an extended period of time and across a variety of areas, one cannot induce something about anything by simply looking at a single factor at a time. Analysing different data over time will give insight into what the problem is, as well as the possible solution. Someone then needs to be empowered to go in and make changes.
Data availability can be a challenge, however. Gathering data that is segregated across email, PDFs, databases, etc, can be time-consuming – data preparation constitutes 70% to 80% of the total effort. However, once all the data has been collected and the variables are in place, an algorithm can be applied in an hour, which will help produce the patterns.
The next step is to measure the effects of the solution that was implemented. If you don’t compare numbers after implementation, you will not be able to quantify the benefits you have achieved.
Either way, analysis should be used for the greater good, from reducing crime, to identifying areas at risk of flooding, allowing time to relocate people before a storm hits.
Goran Dragosavac, SAS Analytical Practitioner