A computer implemented method and system for analyzing a set of data records is provided. Each data record comprises a value for each of a plurality of variables and the method comprises receiving a selection of a variable of interest from the plurality of variables; analyzing values in the data records of at least one variable of the plurality of variables other than the variable of interest; and using the analysis to create a statistical description of said at least one variable. The statistical description is then used to populate at least one template natural language description of a relationship between the variable of interest, or a value or range of values of the variable of interest, and the variable to which the statistical description applies.
