The types of inputs that are commonly used to generate a Lorenz curve and Gini coefficient are non-aggregate variables.
The most common input for the Lorenz curve and Gini coefficient is individual or household income, because income is an important measure of well-being. Other variables could also be the basis for distributional comparisons. These should be non-categorical quantities, such as the amount of water consumed, carbon emitted, or taxes paid. At a higher level of aggregation it also makes sense to consider percentages, such as a fraction of a certain area whose residents are ethnic minorities, voted Independent, or have incomes below poverty. Finally, it may be helpful to consider that the normative implication of a Gini coefficient is that perfect equality is preferred. Certain variables which can be analyzed for their distribution, such as miles of light rail track, may not necessarily be best when equally spread but rather more valuable when grouped together in certain places.
The typical unit of analysis for a Gini coefficient is the individual. UrbanSim allows for the use of the Gini in assessments of equity across units such as gridcells and zones. Parcel level analysis will be supported in the future. It is valid to measure distributions across these units, however the user should keep the units in mind when interpreting the results.