Modelos estadísticos para la detección de patrones en medio ambiente y finanzas :

Abstract

One of the sectors on the rise in modern times is the publicity and advertisements sector, for a written media publicity constitutes one of their main incomes, as such is convenient to conduct an analysis from a data driven perspective, that is, to study data collected by the media and take decisions based on such information. One of these decisions is the election of a main thematic of the media, in the case of a mechanics magazine the thematic of the editions can be from cars to planes including motorcycles and other vehicles and machines. This election is vital for a correct development of the magazine since businesses that work in business lines related will look at the magazine as a platform in which to show the public their product. Therefore we seek to analyze the advertisers from an approach that comprises important characteristics not only of the advertiser company but also of the advertisement that is published. Diverse characteristics of the businesses are taken in account including their business lines, taxes, utilities, profits, assets value and the paid value that covered the publishing expenses. Generating a segmentation of the advertisers using the clustering algorithm “U-k-means” it is possible to give the media a method to plan their editions while having estimates of their possible income based on previous revenue. A data base is constituted from the characteristics of the advertisements and is cleaned and prepared to improve the accuracy and efficacy of the algorithm, generating useful characterizations of the different advertisements groups.