File formats
Choosing an appropriate file format is crucial. Whether during or after the research, it ensures that data remain accessible, interoperable, and reusable, in accordance with the FAIR principles.
Two types of formats can be distinguished:
- Open formats are characterized by publicly available specifications. They can therefore be read, used and, where applicable, modified without relying exclusively on a particular application or provider. They facilitate long-term access to data, reduce the risks associated with software obsolescence, and generally provide better guarantees of long-term preservation.
- Proprietary, or closed, formats are controlled by a specific publisher or provider. They are often associated with a particular software application, instrument, or technical environment. Raw data produced by certain equipment, such as scanners, cameras, recorders, or specialized software, may be generated in this type of format. When the software or environment required to read them is no longer available, using these files can become difficult or even impossible.
It is therefore recommended to save your data in an open format whenever possible.
Recommended practices
The following table lists the recommended and acceptable formats for various types of data.
Please note that this list is not exhaustive and only includes the most common extensions. The format corresponding to these extensions is available here.
| Type of data | Recommended formats | Acceptable formats | Not recommended |
|---|---|---|---|
|
Qualitative data; Texts |
.txt
.xml .pdf/A .odt .odm .tex .md .htm .odf |
.html
.docx .pptx .rtf .eps |
.doc
.ppt .dvi .ps |
|
Quantitative data; Tabular data with extensive metadata Data comprising variable labels, labelling codes and defined missing values, in addition to the data matrix. |
.csv
.por
.ddi .xml .hdf5 delimited text and command file (SPSS, Stata, SAS, etc.) |
.txt
.html .tex .sav .dta .mdb / .accdb |
|
|
Quantitative data; Tabular data with minimal metadata Data that may or may not include column headers or variable names, but which contain no other metadata or annotations. |
.csv
.tab .ods .sql .tsv |
.txt |
.xls / .xlsb
|
| Image data |
.tif (6.0 uncompressed)
.png .svg .jpeg .fits |
.jpg / .jp2
.gif .tif / .tiff .raw .psd .bmp |
.indd
.ait .psd .spc .lif |
| Audio data |
.wav
.flac .ogg .mxl .midi |
.mp3 |
|
| Video data |
.mp4
.mj2 .avi .mkv |
.avchd
.webem .ogv .ogg |
.wmv
.mov .qt |
|
Geospatial data Vector and raster data |
.shp
.shx .dbf Tabular GIS attribute data .gml .netcdf .postgis .geojson |
.mdb
.mif .kml .ai .dxf .svg binary formats of GIS and CAD packages |
|
| Structures and 3D images |
.x3D
.x3DV .x3DB .pdf3D .pov .pdbml |
.dwg
.dxf .pdb |
.pxp
|
| Code |
.r
.py .ijpnb .rstudio .rmd .netcdf .aiml |
.sdd
|
.mat |
| Documentation and scripts |
.rtf
.pdf/A .pdf/UA .xhtml .html .odt |
.txt |
(Source: UK Data Service and EPFL FastGuide #4 - File Formats)
Conversion Tools
There are various ways to convert a file from one format to another, for example, from a proprietary format to an open format.
When conversion results in an alteration of the document’s content or structure, it is advisable, as a precaution, to retain the data in both formats.
Compression Tools
A large volume of data can result in high storage and archiving costs. Compression can help reduce file sizes or combine multiple files into a single archive.
There are two types of compression:
- Lossless compression, which reduces file size without removing any information;
- Lossy compression, which reduces file size further but removes some information.
As with conversion tools, some compression tools can be installed locally, while others are available online.
To learn more
Ressources UNIGE
Capsule Data Science for all - Preparing your data for archiving
Training at UNIGE
-
Organize and name your files efficiently (link to come)
External resources
- E-learning about Open formats: “Choosing the right format for open data” ;
-
Research Data Management Fast Guide #04, "File Formats" from EPFL