In psolymos/QPAD: QPAD estimates

Type of event (training, short course, or workshop)

The type of the event is Short Course.

Title of event

The title of the event is Introduction to the analysis of messy data.

Desired times and dates (up to 3 options)

The proposed event is a full day course (8 hours long with 3 breaks), Monday, August 15 preferred (Tuesday, August 16 less preferred).

Name and affiliation of instructor(s)

Dr. Peter Solymos \ Alberta Biodiversity Monitoring Institute and the Boreal Avian Modelling Project \ Department of Biological Sciences, CW 405, Biological Sciences Building \ University of Alberta, Edmonton, Alberta, T6G 2E9, Canada \ Phone: [phone] \ Fax: [phone] \ E-mail: [e-mail] \ Web: http://peter.solymos.org

Number of participants (min, max) and target audience

Number of participants: minimum 10, maximum 30.

Proposed cost to participants (if applicable)

The instructor is not charging fees for teaching the course.

Summary of course objectives

This course is aimed towards ornithologists analyzing field observations, who are often faced by data heterogeneities due to field sampling protocols changing from one project to another, or through time over the lifespan of projects. Such heterogeneities can bias analyses when data sets are integrated inadequately (Matsuoka et al. 2012, Auk 129:268--282; Solymos et al. 2013, Methods in Ecology and Evolution 4:1047--1058), or can lead to information loss when filtered and standardized to common standards (Matsuoka et al. 2014, Condor 116:599--608). Accounting for these issues is still important for estimating status and trend for bird species and communities, thus better informing large scale conservation and management (Barker et al. 2015, Wildlife Society Bulletin 39:480--487).

Analysts of big and messy data sets need to feel comfortable with manipulating the data, need a full understanding the mechanics of the models being used (i.e. critically interpreting the results and acknowledging assumptions and limitations), and should be able to make informed choices when faced with methodological challenges (Solymos et al. 2012, Environmetrics 23:197--205; Solymos & Lele in press, Methods in Ecology and Evolution).

The course emphasizes critical thinking and active learning. Participants will be asked to take part in the analysis: first hand analytics experience from start to finish. We will use publicly available data sets to demonstrate the manipulation and analysis of large scale and often messy data sets. We will use examples and case studies from the peer-reviewed literature, including the references cited in this proposal. We will teach the use of the following freely available and open-source R packages:

• mefa4 for manipulating large data,
• ResourceSelection package for presence-only data.
• detect for time-removal sampling, distance sampling, single visit based N-mixture and occupancy models,
• unmarked for multiple-visit based occupancy and abundance models.

The expected outcome of the course is a solid foundation for further professional development via increased confidence in applying these methods for field observations.

Overview of topics or syllabus

1. Database relations for field observations (human and recorder based):
2. naming conventions and attributes (survey, session, site, project, visit)
3. geospatial summaries and standard database operations (filtering, sorting, mapping/joins)
4. Estimating nuisance variables:
5. Distance sampling:
   • effective detection radius and effective sampling area
   • constant and covariate models
   • integrating conventional point counts and recordings (ARU based data)
6. Time-removal sampling:
   • constant, time varying, finite mixture models
   • nonlinear responses
   • ARU specific consideration, monitoring changes in behavior
7. Estimating occupancy, abundance, and density:
8. Analyzing presence-only data
   • Resource selection (probability) functions
   • relationships with occupancy and count models
9. Multiple- and single-visit based abundance models and their assumptions
   • Standard and generalized multiple-visit N-mixtures
   • Single-visit N-mixture
   • Generalized distance sampling through integrated likelihood
   • Advantages of conditional maximum likelihood estimation
   • Model selection considerations with mixture models
10. Large scale analysis of integrated data sets:
11. Detectability offsets (QPAD approach)
12. Propagating uncertainty
13. Issues with roadside surveys
14. Computational aspects (high performance computing resources)

Prior experience teaching proposed topic

Dr. Solymos is a statistical ecologist and author of several R packages (dclone, detect, ResourceSelection, mefa4, vegan) with extensive teaching experience in classroom settings and consulting experience including NGOs, governments, academic and industry sectors.

• 1-day workshop 8/2015: Hierarchical models for conservation biologists made easy, ICCB/ECCB congress, Montpellier, France (with Dr. Subhash Lele)
• Bi-weekly webinars (4-6 events / year) in 2014 and 2015 for research collaborators: Manipulating and analyzing continental scale data sets.

Room set up, location and other requirements

• For the course we require a classroom, power for participants' laptops, flip chart OR whiteboard OR overhead projector, internet access.
• Participants should bring their own laptops.
• The instructor will provide open source and free software, and electronic course material online and on DVD/USB sticks.

psolymos/QPAD documentation built on May 26, 2019, 11:31 a.m.