The present article analyzes and explains one of the most current scientific fields— the modeling of phenomena by means of Artificial Intelligence or Artificial Neural Nets. In this case, a traditional demand model of great importance for business marketing decisions is developed. The model is estimated by means of Minimum Quadratic Regression techniques and Artificial Neural Nets. After presenting both the methods for the estimation, explanation and prediction of the demand, these techniques are critically analyzed and complementary research lines are proposed to support the marketing decision-making process.

A model is the specification of a series of variables and their interrelations designed in order to represent a real system. Traditionally, marketing models have been classified according to their purpose and structure (Lilien and Kotler, 1990). Classification according to the purpose is centered on the description and prediction of the marketing phenomena. Marketing decision-making is a less risky task with an explanation of the relationship between a series of variables and the ability to foresee what will happen if one of those variables is modified.

According to their structure, models can be verbal, graphic and mathematical. There is a natural tendency to relate a model with a formal mathematical expression and though this is desirable, most models known today are based on the theoretical verbal formulation of a certain system. Graphic models are an intermediate step, which helps to represent verbal content visually. In this way, the understanding of the model content is simplified and can be studied more easily. However, a few marketing decisions today are based on models. Intuition, inspiration, “sense of smell” and misunderstood experience1 have been, and still are, the prevailing models to explain reality in a lot of companies.