It would have been difficult for most people to predict the exact outcome of the recent budget showdown. While the past history of budget battles may have provided some hints, the final outcome was unknown until the very end.
No one can foretell the future with certainty. However, careful analysis of the past combined with logical reasoning can produce a range of potential outcomes. One technique in this approach is using predictive analytics, which is just a fancy name for statistics. Yes, plain old dull statistics.
Predictive Analytics: Driven by a Clear Business Need
Many management consulting companies and software vendors have been touting predictive analytics as the panacea that will solve all problems. As business intelligence consultants however, we have been advising our clients to view predictive analytics as another tool to use to help solve their business problems. The use of predictive analytics should be driven by a clear business need, either to solve a business problem or achieve a business result. Start with the business problem first and then determine which approach will best help you solve it. In many cases, predictive analytics can help.
Creating the Prediction Model
The technical process for creating a predictive analytics model is fairly straightforward. After doing the precursor step of ensuring a clear business need, follow the process below:
- Formulate the model: a statistician or other specialist will formulate the statistical model to produce the predictive analytics.
- Collect the data: if the required data is not currently in your data warehouse, then you will have to extract the data from the relevant sources.
- Validate the model: this is an iterative process to ensure statistical validity.
- Use the predictive analytics to solve you business problem or achieve a business result.
Now let’s see an example of using predictive analytics at a federal agency.
Analytics in Action: One Agency’s Use of Predictive Analytics
Assume your federal agency delivers citizen-facing services and you count how many units of a particular work item are received and processed at each individual office. You need to become more productive due to an increasing workload and ever diminishing budget. By having a projection of anticipated work over a three, six and twelve month horizon, you will be able to better allocate resources to process this work. Thus you want to predict the number of work items received in your office over the next year.
Following the process outlined above, you create a time series forecast and the predictive analytic “Projected Count.” You then modify your business process to incorporate using this new metric. Overtime, you review the accuracy of the forecast and fine tune the model as appropriate. More importantly, you review if the predictive analytic is helping you achieve your desired business outcome.
Moving Forward with Analytics
Predictive analytics is just another tool for government managers to use in running their organization. With a clear understanding of the desired business outcomes and a properly specified model, you can indeed “predict” the future.