The forecasted demand for that date is generated by assessing
various characteristics of the observed portion of the booking curve, e.g. volume and
shape, and extrapolating into the future date of stay (see, for example, Chen and
Kachani, 2007; Phumchusri and Mongkolkul, 2012, Zakhary et al., 2008, 2009). Two
fundamentally different methods can be used. The first is simply fitting a predetermined
function to the observed portion of the booking curve, and using the fitted function to
estimate the occupancy on the future date of stay. The second method compares the
observed portion of the booking curve to past “completed” booking curves; that is, to
booking curves for past days of stay. Once similar past booking curves are identified
and categorized, their “performance” characteristics are used to predict the
“performance” (i.e. the occupancy on the future date of stay) of the current, incomplete
booking curve