In a data analysis system for determining a correlation model between
biological conditions and expression levels of a plurality of genes
and/or quantities of intracellular substances, A data set is constructed
by using the biological conditions or changes of the biological
conditions probabilistically generated with time as object variables, and
the expression levels of the plural genes and/or the quantities of
intracellular substances as explanatory variables. The explanatory
variables contained in the data are selected, and cross validation is
calculated for a correlation model containing the selected explanatory
variables and object variables to assess the results. Selection of the
explanatory variables, calculation of cross validation and discriminative
assessment of the result is repeated until no further improvement of the
cross validation is observed to determine a partial least square model.
This permits an effective processing method to provide multivariable gene
information.
