Using CBR as Design Methodology for Developing Adaptable Decision Support Systems - Efficient Decision Support Systems: Practice and Challenges from Current to Future - page 129

Subsystem one : each lot of animals consumes several types of feed during each growth stage.
The objective of this phase is to identify the variants of feed that a given lot is going to
consume before slaughter, taking into consideration similar past cases belonging to different
farms and periods. The current output (qualitative information) serves as the basis for the
next phase to compute the total amount of feed needed by both a given lot and a given feed
type (quantitative information).
Subsystem two : given a new lot and the different types of feed that it will consume before
slaughter, it estimates the total consumption (kilograms) by feed type together with the
consumption percentage per animal. Several variables influence the output of this phase:
number of animals by sex (male and female), week of year, mortality indexes, etc.
Subsystem three : this integrates previous calculations by using an ad hoc planning algorithm
for simulating feed consumption and production along with the time. It allows the user to
easily visualize production activity though an intuitive and configurable GUI ( Graphic User
Interface ). Incidences in the initial estimated MPS can be reported and accounted for in
reconsidering daily production.
The data used for constructed the system case base was provided by a leading Spanish
company specialized in animal feed production and delivery. Raw data (from the years 2007
and 2008) was built from client orders, company production logs, information about the
number of animals at different farms and truck trips to the clients. A total of 5112 records
were stored in the SP4 internal case base after executing the initial operation for detecting
and solving existing inconsistencies (procedure 1.A. in Figure 3). In order to increase the
confidence of experimental findings, a cross-validation process was carried out in the all of
the experiments.
Fig. 4. Screenshot of the final deployed SP4 system showing the Master Production Plan