What is the difference between the conventional lift metric and the weighted lift metric used in ProWAR?

Conventional lift ranks association rules purely on statistical co-occurrence, measuring how much more often items appear together than expected by chance. ProWAR's weighted lift, shown as Revenue Lift Optimized, adds profit margins and quantity weights through the Revenue Index, so rules that link high-margin, high-demand items are promoted even if their raw lift is not the highest.

In standard market basket analysis, lift is computed from transaction frequencies only; it describes the statistical strength of an association but does not consider whether the products are profitable. ProWAR was introduced because frequency-based metrics can identify product linkages while missing financial goals such as profitability and demand. In ProWAR's weighted framework, each item is given a Quantity Weight (its share of total items sold) and a Profit Margin, which combine into a Revenue Index (Quantity Weight × Profit Margin). This revenue information is used in the hybrid scoring formula Score = w1 * Normalized Lift + w2 * Normalized Revenue, which produces the revenue-optimized lift values presented in ProWAR's experiments. Conceptually, conventional lift says 'these products tend to be bought together', while ProWAR's weighted lift says 'these products tend to be bought together, and this co-occurrence is financially valuable'. The paper reports that this weighting improved revenue impact compared with traditional ARM and also surpassed earlier profit-based WARM models.