If it's linear, it can be automated. Think about it. What were you doing by hand 10 years ago that is now accomplished at the push of a button? Campaign management? Predictive modeling? Data management and storage? Do you see a trend? So where are we headed?
Technology forecasters predict that there will be hundreds, if not thousands, of new products entering the market over the next 10 years to handle routine activities. In data mining alone, we have seen incredible changes. Years ago we spent weeks building predictive models by hand. Today there are multiple tools that have streamlined and simplified the process.
So, what does that mean to you? If your competitive edge is based on some linear function, your competition may be able to buy software that accomplishes the same thing within a few years. What can you do to stay competitive? Quit using half a brain!
Your challenge is to create an environment that encourages “whole brain” thinking. Some companies routinely practice “whole brain” thinking to provide cutting edge solutions for customers. Whether we're designing a data warehouse, building a predictive model or developing a strategy, we work through the whole brain approach.
The following credit card modeling challenge illustrates how my company used the whole brain method. It involves a case where we combined several linear tools to produce a non-linear solution.
The situation required that we build a model to predict the payment amount of a credit card customer. Since payment amount is continuous, the standard approach would be to build a linear regression model. However, linear regression works best when the dependent variable is normally distributed with the values clustering around an average central value. In this case the dependent variable (payment amount) had a bimodal distribution. In other words, it had clusters of values in two distinct places.
We looked at a distribution and, as you might expect, more than 80 percent paid either the entire balance or the minimum payment. This made perfect sense but didn't fit into the typical modeling routine. So we did some whole brain thinking and came up with an idea. We built two logistic models that are excellent at predicting binary (yes/no) outcomes. Using all of the data, we built a model that predicted the probability of someone paying the full amount and another one to predict the probability of making the minimum payment. We built a third linear model on the 20 percent that paid neither the full or minimum payment.
To implement the model, we scored every name with all three models. We created a set of rules that said: if the customer had greater than an 80 percent probability of paying the full amount, assign the last statement balance as the predicted payment amount; if the customer had greater than an 80 percent probability of paying the minimum payment, assign the minimum payment as the predicted payment amount; otherwise, assign the value predicted by the linear regression. This method provided a 25 percent improvement over the traditional method.
So, why is it so difficult to use creativity? There are several reasons. First, creativity produces variance and decreases predictability. So if management has a high need for control, it is difficult to encourage creative thinking. Another reason is that tapping into our creativity takes concentration. If our work environment is noisy and distracting, it is difficult to access the right side of the brain. And finally, creative thinking requires some down time to get the juices flowing. You might argue that you do not have the time or it is not cost effective. But creative ideas that lead to small improvements to a marketing campaign can save or make millions.
How can you encourage whole brain thinking? This can be difficult if it requires a drastic change in the company culture. But there are a few things you can do as an analyst or manager to facilitate it for yourself and your staff:
Encourage group discussions where ideas are embraced. Brainstorming is an excellent way to get the creative juices flowing.
Change old habits. Sleep on the opposite side of the bed; brush your teeth with the opposite hand; take a different route to work.
Create a workspace that helps you stay balanced. Play music, fill your office with art, spend a few minutes in silent contemplation each day. It's not a waste of time. It's incubation time for the next million-dollar idea.