Lots of companies have been collecting all kinds of data in their data warehouses for a long time. Many think they are ready to perform predictive analytics because they have so much data. Unfortunately, not all data is good information and good information is only one of necessary components.
Some of the data will be irrelevant to use in predictive analytics. If you recall from basic statistics, the data you are using as model input factors has to have predictive ability regarding the dependent variable (target). The dependent variable is what you are trying to estimate. Unfortunately, much of the data collected isn’t what’s needed for predictive analytics. Next, suppose you have found the right factors in your data collection, you will need a very large amount to achieve high reliability in the outcomes you are trying to predict.
Talent is also critical to developing and supporting predictive analytics. While the complex algebra and matrix algebra is daunting, the software can insulate you from the rigors of advanced mathematics. But, you have to know conceptionally the way you are using the software tool. There are lots of “switches” and methodologies you can apply to best fit your data. You just have to know what they are and under what circumstances they best apply. This skill is very important to building a consistent model which reduces your estimate errors.
Hardware enables fast calculation of complex and large data sets. The software calculations are optimized to use the least possible time to get to optimal solutions. One of the biggest challenges is to decide which input variables are best and the correct mix. While there is a lot of science in this sector, there is a lot of art to it as well. Think of your great aunt making a dish – a pinch of this and a dash of that. It’s not all formula driven.
I am excited about seeing predictive analytics come of age. Statistics has been taught in business schools for a long time, but the amount used in business is comparatively light to what is taught. Predictive analytics brings the statistical tools to the forefront. Companies are looking for that “edge”. Why? To gain what Michael Porter defines as a competitive advantage.
Many companies still have a long way to go with their handling of data. Collecting the right data and structuring it is an essential step to get you to the “Starting Line” of using advanced tools like predictive analytics.
There is a lot of power in predictive analytics. It can create a durable competitive advantage. To stay ahead though, you need to get your thought processes ahead of the curve and properly prepare.
