Where does lift collide?

The component of the aerodynamic force on a body moving through a fluid that acts perpendicular to the oncoming flow, and on an aircraft it is the upward force that supports the weight: as air is turned and accelerated around a wing, a pressure difference builds between the two surfaces and the net force across the flow is the lift. Babinsky's 2003 article How do wings work? in Physics Education explains this pressure difference and the circulation of air around the aerofoil, the basis of the dimensionless lift coefficient and of the Kutta-Joukowski theorem that ties lift to that circulation. The referent is a physical force measured on a moving body, not a statistical ratio comparing observed with expected co-occurrence, and not the extra sales caused by an advertising campaign.[1]

Unit

An upward force on a body in a fluid, in newtons, or as a dimensionless lift coefficient

Recovery means

Not applicable

Measured by

Found in a wind tunnel from a force balance, or computed from the pressure distribution and the circulation around the surface

In data mining, a number attached to an association rule that says how much more often two items appear together than they would if they were unrelated: it is the rate at which the items are observed together divided by the rate expected if each occurred independently, so a value above one marks a positive association and a value of one marks none. Brin, Motwani, Ullman, and Tsur's 1997 paper Dynamic itemset counting and implication rules for market basket data introduced this ratio, which they called interest, to rank rules in the market-basket analysis of what shoppers buy together. The referent is a dimensionless descriptive ratio computed from counts, not an upward force on a wing, and not the increase in sales that a campaign actually causes.[2]

Unit

A ratio of observed to expected co-occurrence, dimensionless, centred on one

Recovery means

Not applicable

Measured by

Computed from transaction counts as the confidence of a rule divided by the overall frequency of its consequent item

In marketing and causal inference, the increase in an outcome such as sales, conversions, or responses that is caused by a treatment such as an advertising campaign or a promotion, measured as the difference between what happened with the treatment and what would have happened without it. Lo's 2002 article The true lift model in the ACM SIGKDD Explorations Newsletter set out how to estimate this incremental effect for individual customers and target those whose behaviour the treatment actually changes, an approach now called uplift modelling. The referent is a causal increment estimated by comparing treated with untreated groups, not a physical force on a wing, and not a descriptive ratio of how often two items are bought together.[3]

Unit

The extra outcome caused by a treatment, as a difference in rates or counts

Recovery means

Not applicable

Measured by

Estimated from a randomised test-and-control comparison, as the treated response rate minus the control response rate