Partial stepwise regression for data mining

A computerized data mining method for automatically determining a prediction model for a dependent data mining variable based on at least one independent data mining variable, said method comprising the following steps:

• a variable replacement step (103) replacing said independent data mining variable with potential values from a global range by a multitude of independent local data mining variables, each independent local data mining variable with potential values from a subrange of said global range;
• an initialization step (104) initializing a current prediction model;
• a looping sequence (105-108) including a first step (106) having substeps of
• determining for every independent local data mining variable not yet reflected in said current prediction model a multitude of partial regression functions, each partial regression function depending only on one of said independent local data mining variables;
• determining for each of said partial regression functions a significance value;
• selecting the most significant partial regression function and the corresponding not yet reflected local data mining variable;
• and
• a second step (107) of adding said most significant partial regression function to said current prediction model and of associating said corresponding local data mining variable with said significance value.