Joint Cutoff Probabilistic Estimation Using Simulation: A Mailing Campaign Application
Frequently, organisations have to face complex situations where decision making is difficult. In these scenarios, several related decisions must be made at a time, which are also bounded by constraints (e.g. inventory/stock limitations, costs, limited res
- PDF / 480,862 Bytes
- 11 Pages / 430 x 660 pts Page_size
- 81 Downloads / 186 Views
. Frequently, organisations have to face complex situations where decision making is difficult. In these scenarios, several related decisions must be made at a time, which are also bounded by constraints (e.g. inventory/stock limitations, costs, limited resources, time schedules, etc). In this paper, we present a new method to make a good global decision when we have such a complex environment with several local interwoven data mining models. In these situations, the best local cutoff for each model is not usually the best cutoff in global terms. We use simulation with Petri nets to obtain better cutoffs for the data mining models. We apply our approach to a frequent problem in customer relationship management (CRM), more specifically, a direct-marketing campaign design where several alternative products have to be offered to the same house list of customers and with usual inventory limitations. We experimentally compare two different methods to obtain the cutoff for the models (one based on merging the prospective customer lists and using the local cutoffs, and the other based on simulation), illustrating that methods which use simulation to adjust model cutoff obtain better results than a more classical analytical method.
1
Introduction
Data mining is becoming more and more useful and popular for decision making. Single decisions can be assisted by data mining models, which are previously learned from data. Data records previous decisions proved good or bad either by an expert or with time. This is the general picture for predictive data mining. The effort (both in research and industry) is then focussed on obtaining the best possible model given the data and the target task. In the end, if the model is accurate, the decisions based on the model will be accurate as well. However, in real situations, organisations and individuals must make several decisions for several given problems. Frequently, these decisions/problems are interwoven with the rest, have to be made in a short period of time, and are accompanied with a series of constraints which are also just an estimation of the
This work has been partially supported by the EU (FEDER) and the Spanish MEC under grant TIN 2007-68093-C02-02, Generalitat Valenciana under grant GV06/301, UPV under grant TAMAT and the Spanish project ”Agreement Technologies” (Consolider Ingenio CSD2007-00022).
H. Yin et al. (Eds.): IDEAL 2007, LNCS 4881, pp. 609–619, 2007. c Springer-Verlag Berlin Heidelberg 2007
610
A. Bella et al.
real constraints. In this typical scenario, making the best local decision for every problem does not give the best global result. This is well-known in engineering and decision making, but only recently acknowledged in data mining. Examples can be found everywhere: we cannot assign the best surgeon to each operation in a hospital, we cannot keep a fruit cargo until their optimal consumption point altogether, we cannot assign the best delivering date for each supplier, or we cannot use the best players for three matches in the same week. In this context, so
Data Loading...
