The starting point and the most important element is the business problem. This problem is the essence of the learning process and its solution is the ultimate aim. In general, key business questions revolve around data sets that are closely related, such as sales, number of clients, telephone calls, promotions, etc. There may also exist assumptions or hypotheses that are based on the customer experience, and these form the basis of future solutions. Once the data have been retrieved and these assumptions and hypotheses have been formulated, we can apply statistical techniques in order to let the data speak, transforming the information provided into useful knowledge.

Following this method we develop models that are able to explain the behavior of both markets and customers. The following step is that of reasoned criticism, taking into account other sources of information and the experience provided by customers. The aim is to get congruent models that fit reasonably well with the actual data.

This iterative loop made of assumptions, data, inference, diagnosis, and new formulation is controlled by an economic balance between the predictive power of our models and the utility it will provide to our customers.

Once we have the explanatory models, we make the forecasts. Following the forecasts and the cost function of the company, the decision-making process begins, and those decisions lead to strategies and actions which generate new data, and the feed-back loop continues...