THIS WORK SHOWS THE EFFECTS OF DISCRETIZING A CONTINUOUS ATTRIBUTE-CLASS WHEN TRANSFORMING A REGRESSION PROBLEM INTO A CLASSIFICATION PROBLEM IN A REAL WORLD SETTING. TO DO THIS A CASE STUDY WAS PERFORMED BASED ON THE DATA AS WELL AS CRITERIA GIVEN BY AN EXPERT. THE MAIN OBJECTIVE WAS TO PREDICT THE PERMEABILITY OF AN OILWELL USING INDUCTION RULES, CONSIDERING THE DEPTH, POROSITY AND PERMEABILITY OF NEIGHBORING OILWELLS. THREE DIFFERENT SCENARIOS BASED ON GEOLOGICAL CONSIDERATIONS ARE PRESENTED. THROUGH THE USE OF STATISTICAL METHODS AND VISUALIZATION TECHNIQUES, AN ANALYSIS WAS DONE BEFOREHAND TO GET A BETTER UNDERSTANDING OF THE DOMAIN. THE DISCRETIZATIONS OF THE PERMEABILITY VALUES WERE CONSIDERED (A) IN A STAND-ALONE MANNER, (B) ACCORDING TO THE CONSIDERATIONS OF AN EXPERT AND (C) IN A HYBRID FORM, THUS EMPHASIZING THE INFLUENCE OF THE DISCRETIZATION ON THE PRECISION OF THE RULES. RESULTS OBTAINED USING TWO SYMBOLIC INDUCTIVE MACHINE LEARNING ALGORITHMS, CN2 AND C4.5 -RULES, ARE ALSO REPORTED.