Geoscience Reference
In-Depth Information
17
Embedded GeoComputation
Publishing Text, Data and Software
in a Reproducible Form
Chris Brunsdon
CONTENTS
Abstract .......................................................................................................................................... 397
17.1 Introduction .......................................................................................................................... 398
17.2 Reproducibility in Research ................................................................................................. 399
17.2.1 Addressing the Problems .......................................................................................... 399
17.2.2 Software Barriers to Reproducibility .......................................................................400
17.3 Literate Programming ..........................................................................................................400
17.3.1 Example of the Use of
Sweave
............................................................................... 401
17.3.2 Geographical Example: Choropleth Mapping ..........................................................402
17.3.3 GC Example: Pycnophylactic Interpolation ............................................................. 403
17.4 Implications for GC ..............................................................................................................404
17.4.1 Dealing with Large and Complex Data Sets ............................................................404
17.4.2 Dealing with Difficult Computational Requirements ............................................... 405
17.4.3 Alternatives to LATEX and R ..................................................................................407
17.5 Achieving Reproducibility in GeoComputation ................................................................... 408
17.5.1 Counter Arguments ..................................................................................................408
17.5.2 Ways Forward ...........................................................................................................409
17.6 Conclusion ............................................................................................................................ 410
Appendix 17A Snippet of
Sweave
............................................................................................ 411
Appendix 17B Use of
Stangle
................................................................................................ 411
References ...................................................................................................................................... 415
ABSTRACT
Reproducibility is a major issue for data analysts. In order to subject published results to scrutiny
by peers, it is important to be able to reproduce the steps carried out to obtain them. Furthermore,
failure to do so can lead to major crises of confidence in the results and arguments published.
However, currently such practices are rarely adopted - possibly not at all within the geographi-
cal and geographical information science literature. Here we consider the idea of
Embedded
GeoComputation
in which the computational steps used to obtain results are embedded within
the documents intended for publication. Examples are given as to how this can be achieved for
geographical data analysis, and the challenges involved in making these approaches standard
practice are considered.
397
