Method and system for automated supervised data analysis

1. A tangible computer-readable medium having stored thereon a plurality of executable instructions that when executed by a processor perform a method for automatically analyzing data and constructing data classification models based on the data, the method comprising:

selecting a best combination of methods from a plurality of classification, predictor selection, and data preparatory methods;

determining a best model that corresponds to one or more best parameters of the classification, predictor selection, and data preparatory methods for the data to be analyzed;

estimating the performance of the best model using new data that was not used in selecting the best combination of methods or in determining the best model;

returning a small set of predictors sufficient for the classification task; and

performing optimal model selection and estimating performance of the data classification models using the N non-overlapping and balanced subsets and a nested cross-validation procedure.

2. The computer-readable medium of claim 1 wherein the selecting a best combination of methods comprises:

selecting the best combination of methods from a plurality of predetermined classification, predictor selection, and data preparatory methods.

3. The computer-readable medium of claim 1 wherein the method requires minimal or no knowledge of data analysis by a user of the method.

4. The computer-readable medium of claim 1 wherein the method requires minimal or no knowledge about the domain where the data comes from by a user of the method.

5. The computer-readable medium of claim 1 wherein the method is executed by a fully automated system.

6. The computer-readable medium of claim 1 wherein the method performs comparable to or better than human experts in a plurality of applications.

7. The computer-readable medium of claim 1 wherein the method uses a pre-selected set of classification, predictor selection, and data preparatory methods.

8. The computer-readable medium of claim 7 wherein using the pre-selected set of classification, predictor selection, and data preparatory method comprises:

using a limited pre-selected set of classification, predictor selection, and data preparatory methods.

9. The computer-readable medium of claim 8 wherein the pre-selected set of classification, predictor selection, and data preparatory methods are selected on the basis of extensive tests in application domains of interest.