Scientific Program > Topic 3 > Session 3G >
 Presentation 3G3. Visual Basic Applications and Spreadsheet for Teaching Estimation of Nonparametric Density and Regression Functions

 Presenter Federico Palacios-Gonzalez (Spain) fpalacio@ugr.es

 Presentation Abstract The aim of this paper is to show how teachers can use Visual Basic language on the spreadsheet for students to get some skills for the use of the non-parametric techniques for density and regression function estimations. A useful didactic tool built on Excel Visual Basic for the teaching and practice of nonparametric kernel methods is supplied here. This is intended for students with some preliminary knowledge on this topic. The firs step to follow is to show "EpanechDensity" and "EpanechRegression", user's functions written in Visual Basic for Excel. These functions in Visual Basic are an easy tool for students to get an intuitive perception on the nonparametric estimation meaning for density functions and regression functions. The didactic use of such functions on the spreadsheet and its graphics let show the students the smoothing parameter meaning, the bias effect, the sample size effect and, in general, the estimators behaviour. The different known kernels and a comparative behaviour of such kernels can also be shown by their respective Visual Basic user's functions. Then the rules of thumb and cross-validation methods using the spreadsheet as a tool once again let students obtain a pseudo-experimental and intuitive perception on the topic. User's functions in Visual Basic can also be built for the bandwidth rules of thumb and for Cross Validation function. The later bandwidth can be minimised using the Solver tool incorporated into the spreadsheet. All these functions can be used in the same way as the implemented functions can. Finally, a Visual Basic Application (VBA) will be incorporated as a "Macro" into the spreadsheet. This allows Excel to make use of an added menu similar to a small Statistical Package specialised in the non-parametric methods. Multidimensional cases will also be considered into such a Macro.