In the geology, when you arrive at the end of the exploration and the exploitation tasks and mining planning are begun, it works with a big number of data. These they are generated in different stages and for different people, that produces a range of shapes in the errors that adds to the common ones, taken place during the collection, the preparation of the sample and in the analytic laboratory. Most can be detected by simple inspection, but in a voluminous chart it can be very slow their elimination or correction. Diverse automatic methods of cleaning and correction of data are known for several types of standardized errors, inclusive the errors have been classified to apply to each type a specific algorithm of correction and cleaning, but it doesn't exist a method that is able to clean or to identify errors or possible errors in a global way, the specificity it is the best work tool to achieve satisfactory results, in the case of multivariate data the task are complicated and consistent approaches are needed that can be applied in simple algorithms, It thought about as objective to create a system of rules that allow to identify the errors and automatically correct them, with simple algorithms. It is presented a group of approaches and methods for detection and correction of errors characteristic of the geology, but it leaves open a methodological window to give a similar treatment to the data of the industrial processes.