Computers are intelligent. It’s amazing the intel that can be derived from even a simple data model. Such processes can unearth a gold mine of useful, actionable insight. Everything from ‘Umbrellas sell well on a Friday’ to ‘social media comments posted at lunchtime get twice the views’ right through to ‘in your Glasgow store, you always have zero sales on a Wednesday after 4pm – does it close early’!

However, these completely made-up scenarios are just a snippet of the useful insight a data model can be used for (See our ‘Data Modelling’ page for more examples!) Of course, any data model is only as good as its creator, and the old adage applies just as well here – ‘put rubbish in, get rubbish out!’ and that brings me to the subject of this week’s post.

Firstly, what is a ‘UDF’ or User-defined-field? Put simply, it’s a manually created field that uses some part of the database to give another piece of data. One very basic example of this is a binary score given to each customer based on orders made over clothing seasons. A top score would be 16, shown was ‘1111’, meaning this customer has ordered something in each of your last 4 seasonal ranges. Now when including some UDF’s within a data model, such as may be created on SPSS for example, or SAS, these UDF’s have a tendency to become self-fulfilling prophecies – for example, customers with a binary score of 16 would rank higher on your ‘order predictive model’ for example, but of course they will! We know that anyway – what we really want and need is our data model to rely on the base data only. By concentrating on the base data, the model can use the granularity of each factor and more accurately give a pleasing result.

We pride ourselves on being able to create the very best data models giving increased ROI and incremental sales – contact us for more information.