MPIS algorithm: maximal-profit itemselection with cross-selling effect

( 0 Votes ) 
Категория: ИКТМ 2015 Просмотров: 486
UDC 621.396
MPIS ALGORITHM: MAXIMAL-PROFIT ITEMSELECTION WITH CROSS-SELLING EFFECT
O.O. Slynko*,student, group 535b
D.A. Bielichenko*,student, group 535a
National Aerospace University named after N.E. Zhukovsky "KhAI"


Nowadays the phenomenon of internet-marketingknown as direct-order marketing is widely used. It is at-home shopping systems, E (electronic)-commerce, direct mail marketing. Direct marketing is playing a broader role for building long-term relationships with the customers (direct relationship marketing). It is a direct channel of ordering and distribution, way of dealing with customers directly, bypassing middlemen.

Naturally, sellers are interested to increase their sales and one of the effective ways to do this is to use the cross-selling algorithms.In the literature of data mining, many different algorithms for association rule mining have been proposed. However, there is relatively little study on how association rules can aid in more specific targets. We focused our study on one of the algorithms for association rules - maximal-profit item selection with cross-selling effect (MPIS) problem.
t maybe simple enough to sort items by their profit and do the selection. However, in this case we would have ignored avery important aspect in market analysis, and that is the cross-selling effect.In MPISalgorithm was proposed a new approach to the problem with the consideration of the loss rule - a kind of association rule to model the cross-selling effect. This rule helps to choose the set of items which gives the maximal profit.
The cross-selling effect arises because there canbe items that do not generate much profit by themselves butthey are the catalysts for the sales of other profitable items.\\
We can show that the MPIS algorithm is suitable for real business applications, because it can be transformed depending on the situation toa quadratic programming problem. If the quadratic programming is not applicable for the observed problem, the modeler can use a heuristic approach. Our case study showed the heuristic algorithm tend to be effective and efficient while dealing with cross-selling effect.

*Scientific supervisor assistant V.OButenko.