The University of Michigan and the Federal Reserve Bank of Cleveland separately report inflation expectation metrics that measure what consumers in the United States expect actual inflation (AINF) to be for different time horizons, including 1-year (UMI1 and UST1). This research investigates the following hypothesis: UMI1/UST1 can predict AINF (with an accuracy of >60%) using historical financial indices of: 30-year fixed rate mortgage average, National Home Price Index, Oil Price, unemployment rate, and money supply growth. Collected data [1990-01-01 to 2022-03-01] partially coincided with the Great Recession and the COVID-19 pandemic. To isolate the affected data, five partitions of Calm1, Chaos1, Calm2, Chaos2, and ALL were created. The first and second pairs contain data for the period “before” and “during” the two events, respectively, and the last partition contains all data. Each partition has UMI1 and UST1 versions. The greedy-reduct for each of the ten training sets was identified. Rules were extracted from the reducts using Rough Sets (RS), Association Analysis (AA), and Learning by Example (LE) approaches. The application of the generalized rules on corresponding test sets reveals that AA and RS rules delivered the worst and best performances with six and zero hypothesis rejections, respectively.