Oracle BI 12c (which is compatible with both the “Oracle Data Integrator” versions of BI Applications from 184.108.40.206+ and also the “Informatica” version 220.127.116.11 ) comes with many improvements over previous releases. One I want to look at here is that of “Advanced Analytics”. There are a number of new analytic features that are built into the product based on the “R” language that allow us to simply perform analysis such as Forecasting, detect Outliers, group related items into clusters, add trendlines, etc.
For example, there is a Forecasting function which allows quite sophisticated forecasting via a number of models such as ARIMA ( Autoregressive Integrated Moving Average ) and also ETS ( Error, Trend, Seasonal ).
Below we show sales data for a couple of years (shown in blue) and use the Forecasting function to forecast spend for the next two years (shown in green) using the ARIMA function.
Toggling this to use the ETS methodology we see a slightly different forecast as we’d expect via a different model, but what i’m really highlighting is that there are of course a number of models that allow us to forecast by utilising prior data and various sensitive parameters that allow us to create a scenario that best fits the purpose to which were are looking to utilise it ( e.g. the forecasting of budgeted spend, the forecasting of absenteeism, etc ).
Identification of statistical outliers is also very important. I’ll leave the discussion of what determines an outlier within a specific dataset for now as concepts such as “Mahalanobis distance” are somewhat statistical in nature, but as an example here we use the new Outlier function to process billed quantity for a specific product category and highlight any outlier in both a table and also in a scatter chart.
Outliers would be very useful in a local authority for numerous reasons, perhaps such as identification of P2P data to ringfence customers that have unusual payment patterns or employees with interesting absences.
As a final example of just some of the different types of analysis that can be performed, here’s a Trendline within some payment data. We can plot the complex payments over a number of months and then apply a Trendline function so that we can clearly see the direction of travel.
Here we can see that it is gently rising, which would be positive. This again would be useful in a local authority to see that reduction in absenteeism is heading in the direction we would want or the speed at which SME’s are paid is improving.