Using CBR as Design Methodology for Developing Adaptable Decision Support Systems - Efficient Decision Support Systems: Practice and Challenges from Current to Future - page 127

conflicts in production (i.e. they can not be treated as one single product) and distribution
(i.e. they can not share the same storage areas in vehicles), ( iii ) deliveries to farmers can be
made in advance, but not late, ( iv ) each farm has a limited storage area leading to inventory
holding costs, ( v ) some vehicles can not access certain farms due to their dimensions, etc.
The problem is difficult since it combines two sub-problems known to be NP-hard: a multi-
product, multi-period production problem and a split delivery periodic vehicle routing
problem (Boudia, 2008).
In this context, a worthwhile objective for the main feed production plant of any cooperative
is to determine, on a daily basis, ( i ) the amount produced for each type of feed, ( ii ) the
quantities of feed distributed to each farmer and ( iii ) the associated delivery trips to
minimize total cost over the horizon (setup, holding and distribution costs). In this line we
present the SP4 ( System for Prediction & Planning of Provision Production ) application, a
decision support system that combines a statistical method (used to calculate the previous
consumption data, mortality indices and feed delivery types), a machine learning method
(M5 Prime and IBk models - used to calculate the total amount of feed consumed by type)
and an ad-hoc algorithm which makes flexible orders for compound feed production
forecasting (Reboiro-Jato et al. 2011).

Attribute Type Description

Farm Id Alphanumeric Farm identifier.

Shed identifier. Each farm may have several sheds. This
value differentiates each shed in a given farm.

Shed Id Numeric

Entry date dd-mm-yyyy Date on which the lot arrives at the farm.

Final date dd-mm-yyyy Date on which the lot is slaughtered.

Feed type Enumeration Identifies the specific variety of feed.

Breed Enumeration Specific breed type of the lot.

This variable indentifies the main role of the lot. There
are five possible values: laying hen raising, breeder
raising, laying hen laying, breeder laying and fattening .

Activity Enumeration

Activity specialization. Each activity may have several
specific variants. Possible values include: broiler, label,
light, country , etc.

Subactivity Enumeration

Class of fatten chicken. Animals used on fatten activity
are classified into several categories like 'A', 'C', 'CA',
'H', 'M', etc. This attribute is only used on lots
belonging to fatten activity.

Type Enumeration

Number of males Numeric Number of male animals the lot.

Number of females Numeric Number of female animals the lot.

Number of
unsexed

Numeric Number of animals for which the sex is unknown.

Season Enumeration Period of the year in which the lot spends the most
days on the farm.
Kilograms Numeric Amount of feed consumed by the lot from its entry to
its slaughter.
Table 1. Variables defining the SP4 internal case base