Computational method for matching (using exact-test statistics) geographical regions (zip codes) at the level of individual participants, which allows for quick, easy, and rigorous comparison of every region in the United States on any number of chosen demographics (as provided by US Census data).
CensusMatchR was originally conceived to address market research needs. Regional market research is often of a sensitive nature - questions relating to a potential product or service cannot be asked of residents of the target market because the questionnaire itself could reveal the intentions of the firm. In order to complete the research, surrogate regions must be identified for surveying. The selection of these regions has traditionally been done either by fiat (e.g., on instinct) or by technically inferior methods, e.g. comparing one or two aggregate measures (average age, average income) on their own, due to technical complexity of a more complete comparison. CensusMatchR presents a new method for selecting surrogate research regions.
Read my explanation of the algorithm here - http://steven.ae/censusmatchr_algo.html.