5. Conclusions
A wrong or biased inventory classification will lead to unnecessary
costs due to the non-homogeneity of classes. Sometimes
decision-makers, due to their experience and knowledge, make
decisions based on their preference; this situation generates
familiarity-biased classifications. Therefore, the aim of this paper
is to detect and correct familiarity bias’ presence in items’ classification
done by inventory management’s experts.
Familiarity bias’ detection is performed with two supervised
classification’s techniques; LAD and k-NN, by using cross validation.
The first evaluation is performed using LAD model through
the software cbmLAD, which is based on multi-class LAD pattern
recognition. The second evaluation is performed to confirm familiarity
bias’ presence by the k-nearest neighbor (k-NN) method
based on Euclidean distance, and it was implemented in R
program.