Table 3 reveals that the most favorable prediction model to explain
the relationships between the overall usability and the UWIS
items is a support vector machine model with a radial basis function
(RBF) for the Kernel type. Following the proposed decision
support system in Section 2, the next step is to calculate the sensitivity
scores for each independent variable in this model (RBFSVM)
by using Eq. (5). Thereafter, combining the reciprocals for
each UWIS item from Table 2 with the sensitivity scores in Eq.
(1), the criticality indices for each item are calculated. The next
step in the proposed DSS is to apply a pseudo-Pareto analysis to
those criticality indices to determine out of ‘‘trivial many” which
‘‘vital few” to focus on in order to improve the WIS usability
(Pareto, 1971). The Pareto chart for this case study is illustrated in
Fig. 3. The reason that the word ‘‘pseudo” is used as a prefix for Pareto
is that in this study around 72% of the usability problems stem from
58% of the causes (14 checklist items out of all 24) (as shownin Fig. 3)