Google Summer of Code 2009 Ideas

From Gretl
Jump to: navigation, search

This page is an open list of ideas regarding Gretl's application to Google Summer of Code 2009.

Contents

Information

Timeline

March 9-13: Mentoring organizations applications.

March 13-17: Google SEO program administrators review organization applications for SEO.

March 18-23: Would-be student participants discuss application ideas with mentoring organizations.

March 23-April 3: Student applications.

Contact

You can obtain more information regarding the proposed projects from the organization administrator or by subscribing to the developers' mailing List:

GSoC 2009 Organization Administrator: A. Talha Yalta, Asst. Professor, Department of Economics, TOBB University of Economics and Technology, Turkey

Gretl Developers Mailing List: gretl-devel



Project Ideas

If you are not a developer but have a good Web Desing project idea, please get in touch with the organization administrator or the developers via the gretl-devel mailing list.

Gretl Database Server

Description: the gretl database server is a reasonably small and self-contained CGI program, written in C, that accepts queries passed by a web server and outputs information about the gretl databases present on the host system. Depending on the query, the output is either HTML for display in a browser, or plain text information for formatting and display within gretl, or a gzipped database for downloading. The project involves enhancing this program on several fronts including:

  • Audit its security and make it proof against DoS attacks.
  • Include a search functionality so that a user can easily find all Gretl databases that include a specific variable or country data.

Requirements: C and fundamental network and database concepts knowledge.

Mentor: Allin Cottrell, Professor and Chair, Department of Economics, Wake Forest University

Function Packages Management

Description: At the moment, Gretl manages function packages from within the program. This is a passing solution when packages are few, however, it is not well suited for more complex contributions. Consequently, this project is regarding the implementation of a function packages management system similar to R's library mechanism. It involves introducing a Gretl command (which can be modelled after R's install.library() procedure) for the downloading and installing of various Gretl packages from an online repository as well as the subsequent creation of the necessary GUI entries in Gretl. In the recent years, there is an increased interest in replicability in research in economics. As a result, the system will also support metadata harvestable from external sources in order to provide a mechanism to easily access or cite Gretl packages that are research outcomes rather than just software. Finally, a searchable GUI front-end facilitating this process and making it possible to examine the available packages in various dimensions will be developed as well.

Requirements: C, GTK.

Mentor: Riccardo "Jack" Lucchetti, Asst. Professor, Department of Economics, University of Ancona, Italy

Library for the File Filters in Gretl

Description: In the past, open source programs grabbed code from each other for the importing of data from different file formats used by common statistical software. For example, Gretl's Stata filter is based on code from the R "foreign" package, while R's Eviews filter is based on Gretl's implementation. Both Gretl and R have their SPSS filters from PSPP. This project involves developing a C library that Gretl and other econometric and statistical programs can link in order to centralize some of these efforts.

Requirements: C.

Mentor: Riccardo "Jack" Lucchetti, Asst. Professor, Department of Economics, University of Ancona, Italy

The NIST Test Suite

Description: The accuracy of statistical computations carried out by computers cannot be taken for granted as shown by studies such as McCullough (2004), Yalta (2008), and Yalta and Jenal (2009) to name a few. One important feature of Gretl that most other statistical packages lack is the NIST linear regression test suite, which automatically assesses the regression results through a series of 11 tests using the reference data sets from the U.S. National Institute of Standards and Technology (NIST). This project is regarding the enhancement of this component by firstly making it modular, secondly by improving the presentation of the results, and thirdly by implementing the remaining 3 NIST test suites: univariate summary statistics, analysis of variance, and nonlinear regression.

Requirements: C, fundamental statistics.

Mentor: A. Talha Yalta, Asst. Professor, Department of Economics, TOBB University of Economics and Technology, Turkey

Confidence Band Plots

Description: Visualization is very important for an econometric program and using its Gnuplot integration, Gretl can already produce a wide variety of high quality graphs useful for not only data analysis but also teaching econometrics. This project involves the implementation of the "confidence band plots" for mean Y and individual Y values" (as in Gujarati's Basic Econometrics 4th edition p.144) in the context of a linear regression model. This functionality, like the existing confidence ellipse plots, is especially helpful while using Gretl in a classroom environment.

Requirements: C, fundamental statistics.

Mentor: A. Talha Yalta, Asst. Professor, Department of Economics, TOBB University of Economics and Technology, Turkey

Personal tools