A topological indices vector of five parameters (
,
P,
w,
l,
s) including threegrades of structural informationwas set up as a molecular descriptor to predict the normal boilingpoint, the density, and the refractiveindex of alkenes with a neural network. The five parameters arethe connection index
, the polarity number
p,
w,
l representing the effect of adouble bond on the properties, and
s distinguishingenantiomers of alkenes.The estimation results show average accuracies of 1.3% withmaximum deviations of 16%.