Analysis Effort

Analysis Effort #

This section describes which metrics and procedures should be reported when performing data analysis. It also aims to propose a standardised way to do that so it is easier to compare and share data among different projects.

The idea is to have sections/descriptions that should be in the methods section of a paper, ensuring research can be reproduced, compared and improved. Moreover, it is also important to try and justify the choices as much as possible, either with citations from similar work or ecological reasons for making such decisions

Effort #

How much data was analysed

Unit: Bytes

How was the data analysed? #

Description of how the data was analysed

Example:

  • Data was manually annotated
  • Acoustic indices were calculated
  • Species presence/absence data was collected
  • Minutes were clustered, etc.

Note: for more information on annotations see this link

Tool #

Which software, version, package, was used to analyse the data. Provide all the required information to make sure the research is reproducible.

Example:

  • Raven was used for annotations;
  • AnalysisPrograms.exe was used for generating indices;
  • Audacity was used for detecting species presence/absence;
  • Kaleidoscope was used for clustering and Raven was used for species' identification in each cluster.

How much effort was put into it? #

How much time did it take to analyse the informed amount of data, in a way, using that tool.

Ideally, this information should be the most precise as possible. However, we understand this might not always be possible or accurate. Therefore, even if it is a rough estimate, or only the computational effort, this will help to provide comparison between methods and inform the development of better tools.

Example:

  • X bytes of data were manually annotated in Y hour of effort.
  • U bytes of data were analysed using AnalysisPrograms.exe to calculate acoustic indices. In a computer with ABC specifications, the analysis took D hours to run.
  • F hours were necessary to go through E bytes of files and collecting bird species presence/absence information.
  • Clustering of Q bytes took P minutes with N clusters being generated. Identification of those clusters took O hours.

What was the analysis protocol? #

Was there protocol used to analyse the data in a systematic way? In this section is important to provide justification for every choice, being that the ecological/biological reason for such choice and/or relevant citation that used similar method for answering related questions.

Example:

  • Species were identified in the first minute of every hour
  • Indices were calculated for every minute of recording;
  • Species presence were collected by analysing 5 minutes every 10 minutes;
  • 60 initial clusters were generated and 5 first minutes of each one were identified.

Who performed the analysis #

Provide information on the person who performed the analysis (ideally ORCID).

Data repository #

If the data or any step of the analysis is available in a repository, provide link/DOI for access.


Any additional information that was no covered in this document is valuable as it could improve usability of the data in different contexts. Information on techniques that were tested and did not work is also extremely valuable along with the information/hypothesis of how/why it failed. A lot of time could be saved if we shared not only the success stories but also the “fiascos”.

If you would like to contribute, check this link

Checklist #

  • Analysis effort
  • Data analysis description
  • Information software/method used
  • How much effort was put into analysis
  • Analysis protocol
    • Description
    • Ecological/biological justification
    • Important references/citations
  • Information on who performed the analysis
  • Data repository information