A Procedure for Extending Input Selection Algorithms to Low Quality Data in Modelling Problems with Application to the Automatic Grading of Uploaded Assignments

Figure 3

Test errors for feature subsets of sizes 10 to 20. The results associated to a random forest feature importance-based selection, applied to the centerpoints of the fuzzy data, are drawn with dashed lines. The extension of this method to fuzzy data, followed by a learning with the same centerpoints for Regression Trees, Neural Networks, Support Vector Machines, and Random Forest, but the whole fuzzy data for NMIC, are drawn with solid lines.